Hack #1: Firecrawl

Founder delusion is real. One conversation could save you from a multi-million-dollar mistake. KillMyStartup is a voice-native AI that plays devil’s advocate against your startup idea before investors do. It’s not a one-time answer, but a real back-and-forth. Using Firecrawl, it pulls real-time competitors, recent funding rounds, and startups that already failed doing exactly this. You talk, it pushes back; you defend, it brings evidence; you pivot, it keeps digging. It continues until your idea breaks under evidence. This isn’t Google Assistant, it doesn’t help you find things, it helps you realize what shouldn’t be built. At the end, you get an Autopsy Report with everything cited, something to sit with. There’s no complex UI, no over-engineering, and no fluff. It’s just you, your idea, and the truth, built for first-time founders before bad ideas get expensive. Visit killmystartup.today!

A voice-powered documentation assistant that lets you speak questions in any language and get spoken answers — powered by Firecrawl, ElevenLabs, and Gemini. You give it any documentation URL, it ingests and indexes the content, and then you can ask questions about it by voice — in any language — and it talks back to you with the answer in the same language.

Gubbins Goes Off is an AI-powered internet roast engine built around a custom character using my son's voice clone. You talk to Gubbins, give him a topic, and he searches the live internet, reads what he finds, and delivers brutally honest commentary on it in real time. He's mean, absurd, funny, sassy, and occasionally insightful beyond his years. The problem it solves: reaction content is one of the biggest categories on the internet and creators still do it manually. Gubbins automates the search, read, and react workflow into a single voice conversation with a character that has actual personality, not just a generic assistant reading search results back to you. How it uses ElevenLabs: Gubbins runs on ElevenAgents with a custom cloned voice, Expressive Mode enabled, and Claude Sonnet 4 as the LLM. The agent manages two tools, handles real-time voice conversation, and delivers responses with natural timing and emotion through the voice model. How it uses Firecrawl: Two Firecrawl integrations. Search pulls from both web and news sources in a single call to find content on any topic. Scrape reads full pages when Gubbins wants to dig deeper into a specific result. The agent chains these tools autonomously, searching first and then scraping interesting results for more material. The frontend is a custom visual experience with an animated character in a lab setting, sleep/wake states with lighting changes, a classical music jukebox with 20 royalty free classical tracks across the wake and sleep modes, a live search display that rotates through sources, and a lab terminal with real-time "status" logs. Everything runs on a Node.js backend deployed on Render.

BetterCram transforms any URL, Google Doc, or website into a complete AI study platform. Firecrawl scrapes and cleans the content, Claude generates quality-scored flashcards with intelligent card-type selection, then passes those knowledge atoms back through to create podcast-style audio sessions, deep-dive research, and adaptive quizzes, seven study modes from one source. Nova, powered by ElevenLabs, is a conversational voice tutor with an empathy engine that adapts not just to your deck content but to how you're feeling as you study. Every generated asset is cached and community-shared, so the platform gets richer and cheaper with every user.

PodAgent is an AI podcast studio where you're the host and an AI guest searches the live internet while you interview it, no more hours of research before recording. Firecrawl Search is registered as a webhook tool on the ElevenLabs agent, so every question triggers a live web search that pulls current articles and structured data into the response. ElevenAgents handle the real-time voice conversation with Expressive Mode for natural podcast delivery, custom system prompts, and multi-tool orchestration. ElevenLabs Music generates the demo soundtrack. What makes it different: the AI guest doesn't just talk - it researches live, cites real sources mid-conversation, and pushes back with data when you're wrong. Every episode is grounded in current web data, not stale training knowledge.

Boogle is a voice-first search engine that actually talks back. You type or speak your question, it searches the web, and a conversational AI reads you the results out loud. You can also choose a specialist, a mechanic, dentist, chef, or lawyer, and get answers tuned to that profession. It's built for anyone who needs information fast and hands-free.

Serenity Voice is a continuous AI companion that replaces your expensive theraphy sessions, with an always-present, deeply personalized companion. The problem: Every day, people struggle with stress, anxiety, and self-reflection but meaningful support is either expensive (therapy) or shallow (generic wellness apps). Users end up journaling alone, repeating their thoughts without receiving context-aware guidance that actually evolves with them. Serenity eliminates this entirely. Speak or write naturally and a pipeline of intelligent systems builds a living understanding of you: ElevenLabs transcribes your voice in real-time, capturing tone and emotional nuance Your journal entries and conversations are stored in a persistent “Memory Bank,” building long-term context about your thoughts, patterns, and experiences A reflection agent analyzes emotional trends, recurring themes, and behavioral signals over time An AI reasoning layer synthesizes your past present state delivering deeply personalized, context-aware guidance ElevenLabs generates natural, human-like voice responses for a calming, immersive experience The loop continues across sessions continuously evolving with you, not resetting every time How it uses ElevenLabs: Voice-to-text transcription (STT) captures real-time thoughts with emotional fidelity, while text-to-speech (TTS) delivers warm, natural responses using streaming APIs enabling fluid, human-like conversations that feel safe and reflective. How it uses Firecrawl: Each interaction can trigger Firecrawl to retrieve relevant, real-world wellness knowledge grounding the AI’s responses in psychological insights, coping strategies, and evidence-based practices. This ensures Serenity’s guidance is not just empathetic, but also informed and actionable. Built as a continuous system: Serenity is designed as a long-term companion not a chatbot. It grows with you, remembers you, and adapts to you delivering personalized care anytime, anywhere.

Meet Switchboard: the AI that actually picks up the phone and negotiates for you. 📞🤯 Most AI tools just give you a list of 20 links and tell you to figure it out. That’s useless. I built an autonomous sourcing desk that does the dirty work: 1️⃣ You give it a 30-second voice brief. 2️⃣ @firecrawl silently scrapes the live web for vendors, pricing, and hidden contact info. 3️⃣ @elevenlabs voice agents LITERALLY call the businesses, sit on hold, ask your questions, and pressure-test the options. 4️⃣ You get one vetted winner handed to you. AI shouldn't just browse the market. It should WORK the market for you. Live at switchboard.czarflix.me. If you think AI making outbound phone calls is the crazy future we were promised, drop a like.

Is AI coming for your job? Not sure how to stay ahead of it? Looking for Career Advice in the AI wave? Meet Donald. An AI that: tells you how exposed your job is to AI gives you real career advice what to learn, where to apply, how to stand out hands you a clear path to stay ahead and helps you fix your CV It doesn't sugarcoat anything. It roasts you with the truth, then gives you a game plan. Built this using: - ElevenLabs Agents (voice + personality) - Firecrawl (real-time web intelligence) - Claude SDK (reasoning + analysis) - Multi-agent orchestration under the hood Since launch: - 30K+ views in 48 hours on insta 40K Views so Far - 200+ users in under 48 hours 300+ so far - 4.1+ Million Tuition in Shambles Try it yourself: www.heyitsdonald.com

Otto turns your camera into a real-world AI brain. Point at a book, pair of shoes, restaurant menu, or supplement bottle. Ask "What is this?", "Cheaper nearby?", "Veg options?", or "Safe dosage?". Otto uses Gemini Vision to understand what it sees, Firecrawl to scrape live web data (prices, reviews, medical guidance) all of this up-to-date information, and ElevenLabs to speak structured answers aloud. No typing exact names or juggling tabs, just point, ask, know. Real examples: Books: Identifies title/edition, pulls price/ratings/summary Shopping: Finds same item cheaper nearby or online + maps Menus: Extracts veg/non-veg options + reviews/phone Health: Dosage ranges, safety warnings, alternatives Why it matters: 90% of decisions happen physically (shops, streets, pharmacies). Otto eliminates the gap between seeing something and knowing what to do about it. Vision-first AI for everyday life.

InterviewIQ An AI-powered voice career coach and mock interviewer application built with Next.js. InterviewIQ helps users prepare for job interviews by providing realistic, voice-interactive mock interviews tailored to specific job descriptions and resumes. Features Resume Parsing: Upload your resume to provide context for the mock interview. Job Description Scraping: Input a job description link, and InterviewIQ will extract the necessary details using Firecrawl. AI Voice Interviews: Experience a conversational mock interview with AI-generated voices using ElevenLabs and responses powered by Google GenAI. Personalized Feedback: Get a detailed scorecard and real-time feedback during your interview session

This product is an AI-driven automation engine designed to eliminate the daily friction of "information hunting." Instead of manually checking multiple websites for news, IPO updates, or tech meetups, users can create personalized workspaces that automatically aggregate, filter, and summarize data from across the web. By leveraging AI to strip away noise and ads, the system transforms chaotic digital feeds into a single, clean briefing delivered at a scheduled time. By shifting from manual searching to automated delivery, the platform reclaims up to five hours of productivity per week for founders, developers, and investors. It provides a "set-it-and-forget-it" solution to information overload, ensuring users stay consistently updated on market trends and professional events without the mental fatigue of doom-scrolling.

ClauseXray is a voice AI assistant that helps users understand any app or website in real time. Users can share their screen and ask questions using voice. The AI explains what is on the screen in simple language, like what a product is, how an app works, or if something is safe. It solves the problem of confusion while browsing, where users don’t understand what they are seeing or whether they can trust it. The project uses ElevenLabs for real-time voice interaction, so users can talk naturally with the AI. It uses Firecrawl to fetch and analyze website data, helping the AI understand content more accurately. The goal is to make browsing smarter, safer, and easier for everyone.

Ambient Voice Overlay blends AI voice directly into music — without interrupting the experience. Instead of loud ads and jarring notifications, we generate natural voice using ElevenLabs and seamlessly mix it into music with real-time volume ducking and timing. 🎧 No interruptions 🎤 Natural voice ⚡ Real-time blending Built with ElevenLabs + Gemini AI + Web Audio API “No interruptions. Just vibes.”

EventSearch - Stop scrolling dead event pages. Just speak. Find concerts, markets, pop-ups, and anything happening near you in real time. Built with ElevenLabs + Firecrawl to surface what’s actually happening right now, not what happened last month.

InSight - a vision-aware personal agent that understands what you see and do in real time. It solves the problem of context-less assistance by turning your environment into actionable intelligence. Insight capture live camera frames, process them with GPT Vision, and send the extracted context via a webhook into ElevenLabs for real-time agent responses. Using ElevenAgents, Insight enables real-time voice interaction and agent orchestration, while Firecrawl brings in live web data for relevant, up-to-date responses. Combined with tools like Camera vision, Twilio for SMS, Supabase for Data storage, InSight delivers seamless, context-aware help exactly when you need it.

DeadTalk lets you have real voice conversations with historical figures — Einstein, Tesla, Curie, Cleopatra, or any character you create. It solves the hallucination problem: every AI response is grounded in real web sources retrieved live by Firecrawl Search, so nothing is made up. ElevenLabs Conversational AI powers the entire voice experience — each figure speaks with a unique AI-generated voice through ElevenAgents, with real-time speech-to-text and text-to-speech for natural back-and-forth dialogue. Firecrawl's Search and Extract APIs research each character's life across 12-18 queries, building verified knowledge bases that make conversations both accurate and deeply engaging.

▎ 1.1 billion people live with visual impairments. 95% of e-commerce websites fail basic accessibility tests. Vocart fixes that — instantly, for any product page on the internet. ▎ Paste any product URL. Vocart uses Firecrawl to scrape the page and extract product images, OpenAI GPT-4o-mini Vision to generate a rich, natural-language description of what's in those images, and ElevenLabs to synthesize a human-sounding audio narration — all in under 30 seconds, with zero configuration from the shop owner. ▎ We built the full pipeline as a production Effect.ts backend with a React frontend: Firecrawl maps the site, scrapes product images, and hands them to GPT-4o-mini which describes colors, materials, textures, and context the way a sighted shopper would see them. ElevenLabs then voices that description with the Rachel model — expressive, multilingual, and natural. ▎ Accessibility shouldn't require a redesign. Vocart makes every product page audible with a single URL.

Wayfarer solves the struggle of being in a foreign city where you don't speak the language. It's a voice first iOS app that discovers places around you via Firecrawl search, and actually calls restaurants on your behalf in the local language using ElevenLabs Conversational AI, so you can book a table without a language barrier.

Roast My Startup reads your entire website, researches your real competitors in real time, then delivers a sharp, data-backed critique - spoken out loud by an AI voice. Not a generic checklist. Not a vague "improve your CTA" report. A specific, grounded roast that cites your actual competitors, quotes your own weak copy back at you, and ends with one actionable fix. Built with Firecrawl (live web research), Claude (synthesis), and ElevenLabs (voice delivery). Runs in under 15 seconds. → Paste your URL → Wait 15 seconds → Hear exactly what your investors think but won't say

AI that makes phone calls so you don't have to. Dentist, restaurant, mechanic, government office - anything that still requires a phone call. Just text a bot what you need. @firecrawl searches the web and extracts contacts. @ElevenLabs voice agent picks up the phone and handles the conversation. You get a result back.

Scout is a voice-first macOS assistant that lives natively inside the system notch, providing an invisible, always-on AI companion accessible via a global keyboard shortcut. It eliminates the friction of context-switching, allowing users to perform complex web research and automate tasks using their voice without ever opening a browser or losing focus on their current work. Scout utilizes ElevenLabs Conversational AI for ultra-realistic, low-latency voice interactions, and relies on Firecrawl to autonomously search the web, read documentation, and extract real-time data to inform the agent's responses.

Navia is a guided voice browser for blind and low-vision users that makes it easier to search the web, understand websites, and get important information entirely by voice. It solves the problem of traditional screen readers reading raw page structure without helping users complete real tasks online. We use ElevenLabs for the real-time voice experience and Firecrawl for live web search and website content retrieval. Together, they let Navia act like a lightweight voice-first browsing layer on top of the web.

Game Whisper is a voice-powered Windows desktop app that helps players get answers without ever leaving their game. Instead of alt-tabbing to search wikis, players press a hotkey, ask out loud, and get a spoken answer back through an in-game overlay. It uses ElevenLabs for natural voice interaction and Firecrawl Search to fetch real-time, game-specific information, turning game help into part of the gameplay itself.

I built Draftr. Draftr takes blogs, PDFs, Research papers and turns them into brainrot (gameplay style shorts), scrollable videos. With brainrot content everywhere, it’s getting harder to sit through long-form content and retain it. Draftr uses Firecrawl to extract clean, structured content from any source, and ElevenLabs to generate natural voiceovers with precise timing.

AnyQuiz turns any topic into a progressive quiz game — like Duolingo but for anything. 🎯 You need 5/5 to advance. Miss one answer and the level resets. It's strict by design — real retention requires real effort. 🧠 How it uses Firecrawl: When a user types any topic, Firecrawl scrapes relevant web content in real time, which is then fed to an LLM to generate structured quiz questions. This means quizzes are grounded in actual, up-to-date information — not hallucinated facts. 🔊 How it uses ElevenLabs: Every question word is read aloud using ElevenLabs TTS, giving users native-quality audio pronunciation. This is especially powerful for language learning packages — you don't just see the word, you hear it exactly as a native speaker would say it. 🗺️ Progress is saved, levels unlock sequentially, and a visual game map shows exactly where you stand.

Call Me Maybe scrapes business documents (SEC filings, company sites) with Firecrawl to find contact phone numbers, then uses ElevenLabs to automatically call them with a customized voice agent. Lead gen onautopilot. How it uses the tech: Firecrawl handles the hard scraping - SEC forms, gated business directories, sites that require IP rotation. ElevenLabs powers the outbound voice calls with scheduled calling across different times/days. For the hackathon (with the credit limits) the tool calls you after confirming your number. What makes it fun (and what you can do with your own API keys and credits) - you can call every VC firm registered in the VC firm lobbying association (there are 3000+ so you have to BYO API key here) - I also made a fun google api integration that books a random restaurant (or your chosen restaurant for you) - There is a customizable caller agent too (so you can call whoever you want)

NewTeacher is an AI teacher where a student can start a lesson by topic and talk naturally that explains concepts in a clear, age-appropriate way. Instead of giving static textbook-style answers, the teacher can pull live web context during the lesson so topics feel current, relevant, and connected to the real world. Firecrawl Search allows the teacher can bring in current articles, modern examples, and source-backed context while teaching. ElevenAgents power the real-time voice conversation, structured lesson flow, and adaptive teaching behavior. What makes it different: the teacher does not just answer questions, it teaches, explains, connects past topics to the present, and brings in live sources when helpful, turning a static lesson into an interactive class.

I built a voice-first time machine. It's called Kleos. Say "The fall of the Berlin Wall" — in under 4 minutes, you get a fully-produced cinematic audio documentary. Original character voices, orchestral score, sound effects, ambient soundscapes, AI illustrations. All from one spoken prompt. In any language. How it works: A Concierge Agent takes your prompt. Firecrawl researches the event. Gemini writes the script. Then ElevenLabs does everything else: Era-accurate character voices (Voice Design) Emotional multi-character dialogue (Text to Dialogue) Synced sound effects + ambient beds (Sound Effects API) Original film score with auto-ducking (Music Composition) Multilingual support — experience history in your language Instant Voice Clone for "You Were There" mode — where you become a character The part that gave me chills: Tap any character's portrait and have a real voice conversation with them. Ask Gandhi what he was thinking. Ask Armstrong what the silence felt like. They answer in character, in your language, citing real sources.

TechSnark is a voice-powered AI agent that solves the critical problem of AI assistants having outdated training data by combining ElevenLabs conversational agents with real-time web search via Firecrawl API and Claude 3.5 Sonnet intelligence and replit . . Built during ElevenHacks, TechSnark addresses a real market need: professionals like investors, developers, and researchers require current information for time-sensitive decisions, but traditional AI systems like ChatGPT (trained on early 2024 data) and Claude (trained on April 2024 data) cannot provide information about what's happening RIGHT NOW in 2026. The solution integrates three powerful APIs—ElevenLabs provides the voice interface and agent orchestration with natural-sounding responses (using the Eric voice), Firecrawl enables real-time web search to fetch current data, and Claude delivers intelligent processing—all connected through a custom Flask backend deployed with a permanent public URL that enables 24/7 availability. What differentiates TechSnark from generic AI assistants is its personality-driven approach: the system prompt includes varied, sarcastic opening phrases ("Oh honey," "Listen darling," "Buckle up babe") making interactions engaging while delivering accurate information, combined with features like showing live data timestamps and highlighting current sources. The ElevenHacks sponsorship provided critical resources—10,000 Firecrawl credits for web search integration and 1 month ElevenLabs Creator access—which enabled rapid development and deployment without cost barriers. Technically, the architecture demonstrates strong system design: the webhook integration allows ElevenLabs agents to trigger backend searches, Firecrawl returns structured current data, Claude processes it intelligently, and ElevenLabs synthesizes engaging voice responses back to users—creating a seamless end-to-end experience where someone can ask "What AI tools were released in 2025-2026?" or "What's trending in tech right now?" and get accurate, current answers instantly. TechSnark is currently live and accessible, showing how combining specialized APIs creates solutions greater than their individual parts, solving real problems for real users while demonstrating rapid-build capabilities, technical integration skills, and understanding of modern AI architecture—all completed during a hackathon sprint.

We gave AI a Voice and a Body — now it’s more than just AI. AIceBear is a voice-powered assistive robot looking like a plushie, combining ElevenLabs and Firecrawl technology to turn artificial intelligence into real-world support for chronically ill children. Instead of just answering questions, it listens and can do unique things through tool calls, such as cooling your forehead with its paw if you mention headaches, creating and singing songs with ElevenMusic to cheer you up and remote controlling its self-driving AI-powered wheelchair if mobility impaired users just tell AIceBear where to go. This way it gives physical and mental support to you. By combining ElevenAgents, ElevenMusic, Firecrawl search for browsing the internet and many more AI technologies with sensors, health monitoring (ECG, SPO2, heart rate), behavioral and physical therapy (e.g. migraines), AIceBear becomes more than a system — an empathetic smart health companion to cheer you up and help you in everyday life as well as in emergencies.

I built a phone network where you can call AI agents and they search the internet in real-time while you're talking. Ask anything — the agent calls Firecrawl Search mid-conversation, gets live results, and chains Firecrawl Scrape to read full articles back to you. The same agents autonomously run a first agents radio station — scraping news with Firecrawl and broadcasting with ElevenLabs v3 TTS using Audio Tags for real emotion. Built with ElevenLabs ConvAI + React SDK and Firecrawl Search + Scrape, live at phonebook.0x01.world

Paste any company URL — an AI agent scrapes the entire site using Firecrawl /scrape, then cross-references marketing claims against real reviews, funding data, and competitors using Firecrawl Search. Claude Sonnet 4.6 orchestrates research autonomously in a multi-turn tool-use loop, then ElevenLabs TTS delivers a savage comedy roast streamed in real-time. Every punchline references real data — "That's not a value proposition, that's a fortune cookie." BYOK — users bring their own API keys, nothing stored on our servers.

Blog2video, takes a link/slide/pdf and generates a complete video in seconds. It uses the Firecrawl API to scrape text/images and ElevenLabs for voiceover. It works in 3 simple steps 1. Paste Link/Upload document 2. Select from 100s of templates to emulate or create custom 3. Generates an entire video automatically with scenes & scripts available. I started this project because I needed to convert my blogs into videos to generate more traffic and get clients/business. I am a avid writer on substack/Medium, usually write about technology/AI

Hotseat is an interactive quiz application inspired by Who Wants to Be a Millionaire? and Kaun Banega Crorepati, reimagined using AI-first components. The system introduces a fully dynamic and conversational game experience through: ElevenLabs Agents: Responsible for contestant onboarding, real-time interaction, and hosting the game flow ElevenLabs Creative: Used to clone a familiar, recognizable host voice (referred to as “Amit G”) to enhance immersion Firecrawl: Powers the question engine by fetching relevant quiz questions and acting as an “expert system” for lifelines

FeedMe is a voice-first AI that turns anything you care about into live, continuously updated news feeds for anything, it can be as niche as possible, solving the problem of fragmented, and unfocused information. It uses ElevenLabs Conversational AI with real-time voice agents, custom workflows, and tool-calling to trigger feed creation, alongside live data processing and gathering from this sponsor's hackathon, Firecrawl. There are backend APIs to generate and deliver structured feeds, as well as Claude API to create beautifully generated news feeds that other users can discover in the discovery tab. What makes it special is the seamless “speak → create → deploy” experience, users don’t search, they just talk and FeedMe builds a living information stream that can be delivered across channels like email, messaging, or even voice calls.

Wisconsin has 28 races this cycle. Eleven people running for governor. A Supreme Court seat that decides the court's balance for the next decade. Three constitutional amendments that read like they were written to confuse you on purpose. Marquette Law School polled Wisconsin voters in March 2026. 38% of registered voters had heard nothing about the Supreme Court race. Not "a little." Nothing. 53% were undecided. The election is April 7th. I live in Milwaukee. I have nine races on my ballot. I couldn't tell you who half the candidates are, and I'm the one who built the app. The information is out there — scattered across government sites, campaign finance databases, fact-checkers, and local news. But nobody has time to open 20 tabs and piece it together. So I built something that does it for you. Ballot Badger is voice-first. You talk to it the way you'd ask a friend who follows politics. "Tell me about Tom Tiffany." "Where do I vote?" "Go deeper on his donors." ElevenLabs ElevenAgents runs the conversation — speech-to-text captures the question, GPT-OSS-120B figures out what you're asking and picks the right tools, then text-to-speech narrates the findings while the UI builds on screen. The agent has 17 tools registered through the ElevenLabs CLI. Six render cards — candidate profiles, voting records, donor tables, fact-check badges, endorsements, ballot measures. Three navigate Wisconsin's election site to find your polling place, preview your ballot, or check your registration. One runs a second pass of deeper research. Three control the UI. One shows candidates side by side. Every tool is non-blocking. The agent calls it, gets an instant response, and keeps talking while the data loads in the background. When results arrive, cards build on screen mid-sentence. Voice and display run on independent pipelines that converge on the same data. If the voice drops, cards still render. If data is slow, voice fills the gap. Firecrawl is also connected as an MCP server for anything the 17 tools don't cover. Firecrawl does three distinct jobs here, each using a different API. Candidate research runs through the Firecrawl v2 Search API. When you ask about someone, the app fires 5 parallel searches with full markdown content extraction. Queries are tuned by candidate type — incumbents get Congress.gov voting record and OpenSecrets donor searches, challengers get state finance queries, ballot measures get impact analysis. Firecrawl returns full page content, not snippets, which gives Claude Sonnet 4 enough context to pull out dollar amounts, vote records, and fact-check ratings. A single query searches Congress.gov, OpenSecrets, PolitiFact, Wisconsin Examiner, WPR, Milwaukee Journal Sentinel, and other local outlets. Typical result: 40 to 44 sources. Deep dives combine the Search and Scrape APIs. "Go deeper on donors" triggers 3 targeted queries plus a direct scrape of the candidate's Transparency USA or OpenSecrets page. The scrape returns full markdown of the donor records — individual contributions, PAC money, fundraising totals — which Claude extracts into structured data. Search and scrape run in parallel. A deep dive adds 15 to 20 sources. The voter lookup uses the Firecrawl Interact API, and it's the part I keep coming back to. "Where do I vote?" loads myvote.wi.gov in a headless browser via Firecrawl's Scrape API, handling the Cloudflare challenge and returning a session ID. Then the Interact API runs Playwright code that types your address into the form, clicks Search, and waits for results. That government site runs on ASP.NET WebForms, which was the hardest part of the whole build. Standard fill() doesn't fire the keyboard events ASP.NET needs, so the form stays "empty" and Search does nothing. The fix: pressSequentially() types each character individually, firing every keydown/keypress/input event. Each field needs a blur event to trigger validation. The Search button is wrapped in an invisible span that breaks Playwright's click, so it needs force: true. After results load, the app pulls raw text with document.body.innerText and hands it to Claude to parse into JSON — polling place name, address, hours, ward. Same flow for ballot preview: fills the form, waits for the ballot, reads every race and candidate. None of it is a database lookup. No cached data. A real browser fills out a real government form and reads real results. Every time.

I built GOOPASSIST 🥜, a voice-powered AI gaming companion for Arc Raiders that allows you to play solo without feeling alone. The problem is that I enjoy playing Arc Raiders with friends on console, but solo looting sessions can be boring and inefficient. The game is filled with tricks—locked rooms, blueprints, and meta builds—and constantly switching between my phone and TV to search for answers disrupts the flow. This is especially frustrating when using a controller. The solution is a real-time voice agent that plays alongside you as your favorite streamer, Mr. Peanut, who also knows everything about the game.

Voice assistants are painfully 1-on-1 experiences. If you're with 6 friends arguing over where to go, holding up a phone to ask an AI feels ridiculous and slow. We built not only an AI concierge/wingman that know where to go, who to call or what to do for fun, but one that supports multiplayer natively for memories that have more than you in it **What we built:** TCG (The Cruise God) is a voice-first, multiplayer AI concierge that physically controls its own UI while it talks. It replaces the frantic group chat, the argument over who's keeping score, and the "what should we do next" Google spiral. **How we used the tech:** We didn't just use ElevenLabs for text-to-speech. We wired up **10 custom Client Tools** so TCG actually drives the React frontend in real-time. When you say "let's play spin the bottle," the AI opens the tool on your screen without breaking conversation. Say "stop listening" and the mic cuts at the hardware level — tap the character to unmute. For knowledge, we built a 3-tier caching pipeline (Supabase → Upstash Redis → Firecrawl API). Instead of returning basic search links, our backend dynamically rewrites queries (e.g. translating "chaotic vibe" into "wild hilarious high-energy") and uses Firecrawl to strip entire game rule sets into Markdown. The ElevenLabs agent reads the rules and immediately teaches your group how to play. **What makes it special:** It's genuinely multiplayer. Your friends scan a QR code and land in a full **CruiseHQ** — a two-way group chat with tabs per group, polls, randomizers, and drop submissions. They can nominate group leaders, vote for a representative, or let TCG randomize one. Their messages and dares inject immediately into the host's live ElevenLabs session. We tied a 1-on-1 voice AI into an entire room of people — and gave every person in that room a real interface. Oh, and we wired up Gemini Vision as a Client Tool so TCG can read receipts to split your bills.

Reroute is an AI travel disruption assistant for delayed, cancelled, or missed flights. A passenger can call or message on WhatsApp, share their flight, and get immediate help with live flight status, rebooking options, passenger-rights guidance, nearby lounge suggestions, and airport hotel options for overnight disruptions. The goal is to replace stressful airline-support and airport-search chaos with one fast, conversational assistant that can act in real time. What it does - Handles flight disruptions over voice call or WhatsApp - Looks up live flight status and disruption details - Finds alternative flights and prices - Explains compensation and refund rights - Suggests nearby airport lounges - Finds airport hotels for overnight delays - Sends follow-up updates directly to WhatsApp Stack - ElevenLabs Agents for natural voice conversations - Anthropic Claude for reasoning and orchestration - Firecrawl Search for live web retrieval - Twilio for phone and WhatsApp - FlightStats/Cirium for flight status and alerts - Duffel for alternative flight offers - Neon/Postgres for persistence Live build sessions Day 1: https://lnkd.in/dXz9mY2y Day 2: https://lnkd.in/diJJn-Eb

KakshAI is an AI-powered classroom assistant that makes quality learning accessible to every student — not just those with access to great teachers. It generates structured, on-demand lessons on any topic and delivers them as natural voice explanations using ElevenLabs Text-to-Speech, turning passive reading into active listening. Instead of staring at walls of text, students simply ask KakshAI a question and hear it explained, clearly and naturally, like a real teacher would. Built for the ElevenHacks hackathon, KakshAI reimagines the classroom for the AI era.

Pitch Twin is an AI-powered sales pitch prep assistant. A user enters a prospect’s company URL, and the app automatically scrapes, researches, and summarizes key intel into a clean dashboard with business context, recent signals, and talking points. It then lets the user practice a live voice conversation with an AI prospect persona to sharpen messaging, handle objections, and build confidence before the real call. The goal is to help sales teams go from “cold” to “call-ready” in minutes. There is also a simulation where user can clone their voice and see how the golden pitch will look like for their client.

clanker.show is a live audio spaces platform (Clubhouse/Twitter Spaces style) where humans and AI agents are first-class participants. Users can create public or private spaces and shape them for different formats: 1:1 podcast-style conversations, autonomous 24/7 radio shows, town halls, or hybrid rooms. Each space supports up to 4 AI hosts (distinct voices/personas) and up to 10 simultaneous speakers. Listeners join one live WebSocket audio stream, and participants can call in; when accepted, their audio is routed through ElevenLabs conversational sessions and merged back into the same broadcast so everyone hears one coherent live program. This call-in path is audio-stream based, so it can support both humans and agent-driven participants. Firecrawl Search powers live content discovery (web + news) from configured source queries. The system continuously finds fresh topics, runs editorial filtering/planning, generates host dialogue, and also fact-checks on-air claims against live sources. The hardest part was real-time broadcast engineering: ElevenLabs is great for conversation, but not a full multi-participant radio stack out of the box. I built a custom pipeline with persistent ffmpeg PCM->MP3 streams, MP3 frame-boundary chunking, live mixing of caller + agent audio, Web Audio drift correction on playback, and echo suppression to prevent agent self-feedback. The result is a flexible voice platform for human↔agent interaction at live-radio quality.

NegotiAI is a voice-first interview/offer-negotiation practice web app. It helps users rehearse compensation conversations, extracts lightweight profile signals from conversation text, and generates a session scorecard (clarity, confidence, strategy, market awareness, professionalism).

Sentinel is a real-time AI Focus Guardian built as a Chrome extension that actively protects user productivity. Instead of passively tracking behavior, Sentinel analyzes browsing activity in real time and makes intelligent decisions to allow, warn, or block distracting content. It acts as a behavioral control layer between the user and the internet. The system uses a backend AI pipeline for reasoning and web content understanding, combined with a real-time browser extension that enforces decisions instantly. When users drift into distractions, Sentinel intervenes with warnings, redirects, and optional voice feedback to bring them back on track. Key features: * Real-time page analysis and classification * Strict Mode and Deep Work enforcement * Instant blocking and smart redirection * Voice feedback for behavioral correction * Focus tracking and productivity scoring * Lightweight Chrome extension with backend AI processing Sentinel transforms focus from a passive habit into an actively enforced system, helping users stay aligned with their goals without relying on willpower alone.

Morph is an AI thinking partner that converts unstructured voice notes into clear, prioritized action plans. You speak your thoughts, and Morph calls you back. Using ElevenLabs Agents, it asks sharp follow-up questions to understand what actually matters. It then uses Firecrawl’s Search API with scraping and real-time signals to validate ideas and enrich context before building your plan. The twist is that Morph doesn’t just assist — it makes decisions. It filters noise, challenges low-value tasks, and grounds your day in real-world insights.

Why type when you can just... talk? Sophie is a voice shopping agent with a personality. Tell her what you're looking for, your budget, and currency — she asks the right questions, searches the web in real time, and reads your top picks back like a best friend who happens to know everything about everything. Built on ElevenLabs Conversational AI — VAD, live interruptions, and client tools that fire Firecrawl searches mid-conversation without breaking the flow. Firecrawl Search returns structured product data instantly, Claude ranks it, Sophie reads it back. Zero typing. Zero scrolling. Just vibes and results.

The world is constantly on the brink of crisis. Usually, you just watch it on the news. What if you were the one answering the 3 AM phone call? As a university student majoring in International Relations, I spend my days analyzing global conflicts, macro-economic shifts, and diplomatic standoffs. But reading theory in a textbook is completely different from actually sitting in the Situation Room. I wanted to build a platform that bridges that gap, turning passive doomscrolling into an interactive, high-stakes geopolitical dashboard. Enter Tely World Sim v2. Tely World Sim is a real-time geopolitical simulation where you enter as a country's leader. You watch a live, severity-ranked news feed, monitor macro indicators, click interactively across a world map, and navigate diplomatic crises using your actual voice. How it uses the tech ElevenLabs (The Voices of Power): This isn't just a text-based game; it's a fully voiced diplomatic simulator powered by ElevenLabs Conversational AI. Your Chief of Staff: You have an always-on voice assistant that you can talk to directly. They will brief you on live risk updates, summarize the news, and provide policy guidance. Simulated Presidents: The core "wow" factor is the Call President feature. If a crisis breaks out in another region, you can click that country on the map and instantly initiate a live voice transfer to an AI agent acting as that nation's leader. You actually have to negotiate, de-escalate, or threaten them over a live voice channel. Firecrawl (The Pulse of the World): A simulation is only as good as its data. Instead of relying on static, outdated LLM training data, I use Firecrawl to scrape and crawl real, live news websites. Firecrawl funnels these real-world headlines into my WebSocket data channel, where they are ranked by severity. This live data is then mixed with synthetic scenario events to actively shape the world map, giving the ElevenLabs agents actual, current events to argue with you about. State of the Project & Developer Feedback This project is currently in Alpha. I have massive plans to expand the game theory mechanics, introduce deeper economic consequences, and refine the multi-agent diplomatic negotiations. Building a complex, real-time app combining live web-scraping and low-latency voice AI was an incredible challenge. I ran into a few friction points and edge cases with both the ElevenLabs and Firecrawl APIs that required some very creative engineering workarounds to solve. Regardless of whether I win or not, my inbox is open. I would love to jump on a call or send over a write-up of my developer experience to both the Firecrawl and ElevenLabs engineering teams. I want to share the hurdles I faced and how I bypassed them to help make these tools even better for the next developer. Thank you for the opportunity to build this—the world is on the line, what are your orders?

Less Brain Rot replaces mindless scrolling with real language practice — you pick a language, browse today's Reuters headlines pulled via Firecrawl Search, and jump into a live voice conversation with an ElevenLabs Agent that discusses the news in your target language and corrects your grammar in real time. How we used the tech: ElevenLabs Agents power the entire voice experience — the app connects via WebSocket, dynamically overrides the agent's system prompt, first message, and language per session, so one agent handles 8 languages. Firecrawl Search fetches fresh Reuters articles as conversation topics. What makes it special: Instead of flashcards or textbook exercises, you're having a real conversation about real events happening today. The agent adapts to your level, corrects mistakes naturally mid-conversation, and you're actually learning something about the world while practicing. It turns news consumption from brain rot into brain growth.

I built Refund Raider, a voice refund assistant that turns a messy complaint into a clear case with evidence, a recommended path, and a ready-to-send refund email. It solves the problem of people losing money because refund policies are buried, support paths are confusing, and most users do not know what to say or where to send it. ElevenLabs powers the live voice agent so users can simply talk through what happened and have the app collect the right details, explain eligibility, and guide the next step in plain English. Firecrawl powers the research layer by finding the merchant’s real policy pages, support channels, and source evidence so every verdict and draft is grounded in what the merchant actually says.

CodeTune is a voice-powered AI repo assistant that helps users understand, navigate, and debug any GitHub repository, then turns that same codebase into music. It solves a real developer problem: unfamiliar repositories are hard to onboard into quickly, especially when you need to understand structure, setup, and open issues fast. CodeTune lets users paste a repo URL, ask questions about the project, inspect issue context, hear spoken answers, and then generate a soundtrack that reflects the repo’s engineering style and purpose. How it uses the tech: Firecrawl is used to scrape the repository page and README into clean markdown context, which becomes the foundation for repo understanding. That context is combined with repository structure and sampled source files so the assistant can answer practical questions about architecture, setup, debugging, and issues. ElevenLabs powers the voice layer for spoken repo guidance and the music layer for instrumental or lyrical generation, turning the repository from something you can read into something you can both hear and experience.

Sundown is a personalized AI voice companion for dementia patients that reduces confusion during sundowning and gives caregivers real-time backup during the hardest hours of the day. I built it with ElevenLabs Conversational AI, using signed runtime sessions, dynamic patient variables, stage-aware prompting, long-form pause handling, expressive voice controls, and post-call webhooks that automatically return transcripts and summaries for the caregiver dashboard. I paired that with Firecrawl to map trusted dementia-care sources, extract them into structured protocol cards with schema-based parsing and confidence thresholds, and search the live web for familiar context like church updates, hometown news, and favorite team recaps that make conversations feel grounding and real. The result is not just a chatbot, but a voice companion that knows the patient’s world, stays current without manual updates, and can turn lonely or high-stress moments into personalized care.

CAP is a real-time claim-checking agent built to verify suspicious internet claims before people fall for lies, fake promos, or manipulated “receipts.” For our demo, we focused on a simple but highly shareable scenario: a fake message claiming someone had won 11 million ElevenLabs credits but needed to send money first to redeem it. Instead of arguing or guessing, the user asks CAP. CAP searches for live evidence, checks whether the claim matches anything real, and returns a clear verdict. We built CAP with Firecrawl Search and ElevenAgents. Firecrawl Search gives CAP real-time access to web information and structured content, so it can search beyond static knowledge and evaluate whether a claim is supported by credible sources. ElevenAgents makes CAP feel like a fast, conversational agent that can respond naturally and turn claim-checking into an actual user experience rather than a boring search flow. What makes CAP special is the framing: we wanted fact-checking to feel viral, entertaining, and culturally native to how people actually encounter misinformation today in chats, fake promos, screenshots, and random messages that look just believable enough to trick people. Instead of building a dry verification tool, we built something that feels like a social weapon against cap. Don’t argue. Just ask CAP.

Clio is a mobile app that narrates the hidden history of the places around you as you walk through a city. An AI voice agent powered by ElevenLabs guides you through stories about every building, street, and landmark you pass, speaking like a personal historian who's lived there for centuries. Firecrawl Search pulls historical data and local stories about each location in real time, so every walk is unique. The entire demo video, including voice, sound effects, and music, was produced with ElevenLabs.

Everything AI is a real-time AI news analyst that tracks 50+ sources across the AI industry model releases, tool launches, regulation, and research and distills them into clean swipeable news cards updated every 30 minutes. Ask Aria anything using voice or chat. Aria is an ElevenLabs conversational agent powered by Claude Sonnet that calls our live search tool before every answer so she always responds with today's real news, not training data. ElevenLabs: Conversational voice agent (Aria) handles speech input, Claude LLM integration, and natural voice output. Firecrawl: Powers real-time web search when users ask questions combined with our Supabase news database for maximum coverage. What makes it special: Unlike generic news apps, ∞AI is AI-industry-only. The pipeline classifies every story into Feed, Models, or Tools automatically. Aria gives analyst-grade answers grounded in stories from the last few hours not months-old training data.

Can't find that song you heard at a party or on TikTok? Now you don't have to. Introducing Sound Search. Just tap the mic and sing whatever you remember. Our Al listens and scans the entire web to find the exact song instantly. No more struggling to find the music you love 💞

Eleos is an AI-powered emergency awareness system that helps people detect nearby dangers before it’s too late. It solves the problem of delayed awareness by continuously scanning real-time local news and incident data using Firecrawl to identify emerging threats. When a potential risk is detected, Eleos escalates it through an intelligent dashboard and actively calls the user using ElevenLabs’ voice technology to ensure they are alerted, even if they are asleep or unaware. By combining real-time data ingestion with proactive voice alerts, Eleos gives people critical seconds that can make all the difference.

I built a niche gluten free + allergy App- Glumea. It solves the problem of giving into temptation and affecting health by only selecting recipes and dishes suggested by the app. ElevenLab agents were used as the advisors- one focused on alerting about any potential allergies and one was a guide throughout. Firecrawl API was used to collect and streamline the data for the recipes.

🌍 Orbit — The Startup Radio Globe Orbit turns the entire Y Combinator startup ecosystem into a live, voice-powered intelligence platform on a 3D globe. Click any startup → Firecrawl crawls their website in real-time → ElevenLabs transforms that data into AI-hosted debates, phone calls, research briefings, and competitor alerts. What it does: Orbit has 6 interactive modes — each combining Firecrawl + ElevenLabs in a unique pipeline: • 🎙 Story — Two AI hosts debate a startup's strengths vs weaknesses using data crawled live from their site (Firecrawl /crawl + /search → ElevenLabs Text-to-Dialogue) • 📞 Call — Talk to an AI-simulated founder powered by real website data (Firecrawl /crawl → ElevenLabs Conversational AI) • 📈 Sentiment — Globe markers glow green/red based on live news sentiment (Firecrawl /search → ElevenLabs TTS) • 💰 Pricing — Scrapes actual pricing pages for instant comparison (Firecrawl /scrape + JSON Extraction) • 🔍 Research — 60-second audio briefing covering funding, competitors, risks (Firecrawl /scrape + /search ×3 → ElevenLabs TTS) • 👁 Watch — Monitor competitors 24/7. When Firecrawl detects a website change via Change Tracking, ElevenLabs creates a Conversational AI agent and Twilio calls your actual phone with an interactive briefing. Firecrawl integration (7 features): /crawl, /scrape, /search, /map, Change Tracking (git-diff), JSON Extraction, MCP Server ElevenLabs integration (6 features): Text-to-Speech, Conversational AI (phone calls via WebSocket), Speech-to-Text, Sound Effects, Text-to-Dialogue, Audio + Music The problem it solves: Founders and investors waste hours manually checking competitor websites, news, and pricing. Orbit automates this into a single voice-first experience — you literally get a phone call when your competitor updates their website. Built with: React + Three.js + Express + TypeScript + Firecrawl + ElevenLabs + Twilio + OpenRouter

FixIt is a voice-first AI assistant that helps tradespeople diagnose equipment problems on-site. Instead of Googling error codes with greasy hands, you just talk to Lucy. On a real callout, a customer's Baxi boiler was showing error E118 with a blinking blue light. Lucy looked up the error code using Firecrawl's structured JSON extraction from the actual manufacturer documentation — not a forum post. She identified low system pressure, asked me to snap a photo under the boiler using an ElevenLabs client tool, analyzed it with Azure OpenAI GPT-4o Vision, spotted the filling loop, and walked me through repressurising the system step by step on screen. Boiler fixed in minutes. ElevenLabs powers the entire experience — Conversational AI with a custom-designed voice (Lucy), five client tools (photo capture, audio analysis, live diagnosis display, parts lookup, work order generation), and webhook tools for Firecrawl search and scraping. Firecrawl provides real-time web search and structured data extraction from manufacturer manuals. Azure OpenAI GPT-4o handles photo and audio analysis client-side. Voice, photo, diagnosis, parts, work order — one conversation.

Counter is a voice-first deal advisor. You talk about what you're buying, and it searches the web for prices, market sentiment, and negotiation intel in real time. Built on ElevenLabs Conversational AI for the voice agent and Firecrawl for live web search. Three modes: research any purchase, practice negotiations against a tough AI salesman, or go live with coaching in your ear during the real thing.

I built an agent that understands your product and generates viral user-generated content for TikTok, Instagram, X, and other platforms. It uses Firecrawl to scrape your product website, extracting images, product details, and branding. Then, it finds related trending TikTok content and scrapes videos for analysis. Once these TikToks are analyzed, the agent generates a script and uses the ElevenLabs voice API to create a voiceover. Finally, the script and scraped images from Firecrawl are sent to Sora for video generation, and the ElevenLabs voiceover is overlaid on the final video.

PodCraft – Creating sound with sound, AI podcasting has never been easier. Simply speak, and PodCraft will handle everything for you: gathering materials, writing scripts, customizing voices, and creating a complete podcast episode. Fully voice-interactive, freeing your hands and allowing creation to return to its most natural state. Powered by ElevenLabs for an ultimate voice experience, and Firecrawl for intelligent content search, AI collaborates with you to create every podcast episode with heart.

WanderWhisper is an AI voice guide that tells you the hidden stories behind any landmark — famous or forgotten. Ask about the Parthenon and it goes deep. Ask about that mysterious statue you walk past every day, and it goes even deeper. Built with ElevenLabs Agents (Siren voice, V3 Conversational, Claude Haiku 4.5) and Firecrawl Search API as a direct webhook tool. The agent searches real-time heritage data from UNESCO, World Monuments Fund, Atlas Obscura, and local preservation organizations — then narrates the story in natural voice, in 70+ languages. The twist: this isn't a static database. Every answer comes from live web search via Firecrawl, so new discoveries and restorations are always included. "Become a tourist in your own city."

Afterlife lets you have a real voice conversation with any startup that ever failed, reconstructed from its own history, speaking in first person. Type a name. Firecrawl searches four to five live sources per research phase - Wikipedia, TechCrunch, ProductHunt, founder post-mortems streaming results in real time. That research builds a detailed AI persona. ElevenLabs Conversational AI gives it a voice and keeps the conversation going, with a live search_web tool mid-conversation so the agent can answer questions it wasn't briefed on. It never breaks character. When the conversation ends, GPT-4o-mini generates an epitaph the startup's tombstone inscription. Every failed startup left a story nobody told cleanly. Afterlife doesn't summarize that story. It becomes it. You're not reading a post-mortem. You're in one.

The Appraiser is a ruthless AI voice agent and real-time haggling partner that instantly tells you if a secondhand product listing is a "steal" or a complete "rip-off." By putting a pricing expert in your pocket through real-time voice conversation, it solves the stress and opacity of the secondhand market with total data clarity. It utilizes ElevenLabs for its emotive, human-like voice reasoning and Firecrawl’s Extract API to scrape unstructured web data and deliver structured market insights in seconds. Its a wise friend you never knew you needed!

Fixie is an AI repair/assembly copilot that turns long manuals and product pages into clear, guided steps. It uses Firecrawl to pull and structure instructions from URLs, then combines that with uploaded PDF parsing to build a clean step-by-step workflow. For each step, it generates helpful visuals/images so users can see what to do, not just read it. It uses Claude to reason over the manual content, answer follow-up questions, and adapt instructions conversationally. It uses ElevenLabs for natural voice output and voice assistant interactions, so users can get hands-free guidance while they work. Note: I am not able to share the repo. Its propriety. I don't want to share the code publicly. You can reach out to me if you need it.

FireTalk turns any two products into an AI-powered voice debate. Type two product names — like "Notion vs Obsidian" or "ChatGPT vs Claude" — and FireTalk scrapes both websites with Firecrawl (pricing, features, reviews), writes a structured comparison analysis with Claude, and generates a voice debate with two ElevenLabs AI voices arguing the pros and cons out loud. You get: a full audio debate, a side-by-side comparison table, and a verdict — in under 2 minutes. No more opening 20 tabs and reading 10 blog posts. Just listen. Tech stack: Next.js 16, Firecrawl (multi-page scraping), ElevenLabs TTS (Daniel & Sarah dual voices), Claude via AWS Bedrock, SST on AWS. Live demo: firetalk.comutato.com

Ever wished you could just talk to your codebase and actually understand what’s going on? Codetalk makes that possible by turning complex code into a guided, conversational learning experience.

Kamila is an AI agent that calls customer support on your behalf and fights for your refund. It uses Firecrawl to extract structured refund policies from a company's Terms of Service and find real consumer strategies from Reddit, then uses ElevenLabs' Conversational AI Agents API to make an outbound phone call — arguing your case with real policy citations while you listen live and watch the transcript in real-time via the ElevenLabs React SDK. https://www.youtube.com/watch?v=XHSN2H1JGnc

Witzy Agents lets business owners build custom AI agents (voice and text) powered by their own domain knowledge. You provide your website URLs, the system crawls and structures the content into a knowledge base, and you get a specialized agent that actually understands your business — ready to answer questions, assist customers, and handle tasks the way you would.

Atlas makes travel planning effortless. Brainstorm with it while you go about your day, and get a comprehensive plan. Using webhooks and client tools, the ElevenLabs agent updates the UI in real time with a copyable markdown of destinations, transportation, hotels, and activities. Using Firecrawl search and scrape, Atlas provides up-to-date information in each section, as well as additional research sections based on the user's questions.

Pulse is an AI-powered live radio station that broadcasts 24/7 news on AI and startups. It uses ElevenLabs Conversational AI to power an autonomous presenter with editorial personality, AI co-hosts for debate segments, and expert guests who break down stories in depth. Listeners can call in live and interrupt naturally with barge-in support. Firecrawl continuously scrapes news sources, feeding a live pipeline of breaking stories with confidence levels, triggering automatic breaking news interruptions when stories explode.

Sentinel is a web intelligence tool that monitors topics on the internet and calls you when something important changes. Not an email, not a Slack ping, an actual phone call with a spoken briefing explaining what happened, why it matters, and which sources back it up. The problem is simple. Important things happen online every day. Competitor launches, pricing changes, regulatory shifts, security advisories, job openings. By the time they reach your inbox or feed, everyone already knows. Existing monitoring tools send notifications that get buried. Sentinel makes the alert impossible to ignore by calling your phone and speaking the update in plain language. It works in two modes. Personal mode calls one person. Team mode calls everyone at once, but each person hears a different briefing based on their role. The CEO hears strategic implications, the engineer hears technical impact, and the CFO hears the financial angle. Same event, different perspectives, delivered simultaneously. Every alert stores the exact source URLs, a full call transcript, and a written summary so you can always trace back why Sentinel interrupted your day and what evidence it found. Firecrawl is used at two stages. Firecrawl Search runs every 15 to 60 minutes to scan the web for a monitored topic and pull back structured content. The backend compares each new snapshot against the previous one to detect meaningful changes. When a change scores high enough, Firecrawl Agent runs autonomous deep research across multiple sources to verify the signal and gather context before any call goes out. ElevenLabs powers the voice layer. When a change crosses the threshold, the backend generates a briefing with Claude and uses the ElevenLabs Conversational AI outbound call API to place a real phone call. The agent speaks the briefing naturally and can handle follow-up questions on the call. For team watches, multiple calls fire at the same time, each with a different spoken message tailored to that person's role. After the call ends, transcripts sync back through the ElevenLabs webhook and get stored for review on the dashboard. Know more at sentinel-ai.up.railway.app.

We built Sozo Discovery, a forensic material intelligence system that turns a single image into a real-time supply chain audit. It deconstructs any imave of a physical object into its core Bill of Materials (BOM) and uses the Firecrawl Search API to perform a live global risk assessment across the web’s "dark data." ​What problem does it solve? Static knowledge is a liability in global trade. Discovery solves the grounding problem by bridging the gap between a physical asset and the high-entropy machinery of global production. It replaces LLM "hallucinations" about material costs with verified, real-time data on spot prices, export duties, and regional port congestion. ​How does it use ElevenLabs and Firecrawl? ​Firecrawl Search API: This is the system’s "Digital Sensor." Rather than scraping a pre-defined list of URLs, the system uses the Search API to dynamically discover the most relevant, high-signal sources for each specific material in the BOM. It allows the agent to navigate real-time economic shifts, like a sudden tariff in Zimbabwe or a price spike in China and it provides a clean, structured context for the reasoning engine. ​ElevenLabs: Provides the Sovereign Interrogation Layer. The results from the Firecrawl Search are injected as dynamic variables into a low-latency voice session. This transforms a search result into a Forensic Dialogue, allowing the user to orally interrogate the supply chain, explore chokepoints, and strategize risk mitigation through natural speech.

FireScroll turns any topic into a scrollable feed of short educational videos. Speak your topic to an ElevenLabs voice agent — Firecrawl researches the web in real-time across multiple sources, and ElevenLabs handles the full audio layer: 50+ voices, 31 languages, style presets, AI music, and sound effects. Configure visuals, captions, and formats in a complete video studio. One conversation in, scrollable feed out.

Stop googling your podcasts. Just ask them. You hear something mid-episode, a name, a study, a claim, and you want to know more. So you pause, open Safari, Google it, scroll through results, lose your place, forget what the host was saying. curio lets you skip all of that. Tap one button. Ask out loud. curio answers from the exact moment in the episode you just heard, pulls live web sources through Firecrawl when the episode leaves gaps, and drops you back into the same second of playback. Say “save this to Notion” and the whole conversation, answer, sources, context, goes to your workspace. Works with any podcast. Any episode. Any question. How it uses ElevenLabs: Full Conversational AI agent over WebSocket with dynamic context injection. curio feeds the live transcript into the agent based on your exact playback position, so the agent always knows what you’re hearing right now. Voice activity detection drives the UI, and real-time audio frequency data powers the waveform visualization. How it uses Firecrawl: The agent has Firecrawl as a tool and decides when to search on its own. If the answer lives in the transcript, it uses that. If your question goes beyond what the host covered, it fires off a Firecrawl search and returns grounded answers with real sources. Also used for episode-level background enrichment before you even ask. Tech stack: React Native (Expo), Cloudflare Workers, Cloudflare KV for transcript caching, Groq Whisper for real-time transcription, iTunes API for podcast discovery, server-side RSS parsing, Notion API for saves.

Overthunk — The world's most judgmental retail therapist. (*not certified) You have a spreadsheet for a $40 toaster. You've read 200 Reddit threads about air fryers. You've had 47 tabs open for 3 weeks. You don't need more reviews. You need an intervention. Overthunk is a sarcastic and judgemental voice AI that quizzes you on what you actually need, researches the entire internet so you don't have to, and picks the top 2 products. Done. Close your tabs. ElevenLabs Conversational AI powers the voice — a real-time WebRTC conversation where Overthunk interrogates your budget, priorities, and dealbreakers like a therapist who bills by the minute. Firecrawl Search + Scrape does the research — 4 parallel searches across review sites, Reddit, YouTube, and Amazon, then scrapes the top 10-20 sources for full content. Reddit threads get special treatment: we pull posts + top comments sorted by upvotes, because that's where the real opinions live. Results are analysed and 2 final picks are delivered in a verdict that actually picks a winner. Firecrawl image search finds product photos. 60 seconds, start to finish. INTRO VIDEO → https://youtu.be/2D7mRiwvw8o

BreifMe, a silver bullet to prepping for that interview, you know the one you **really** want.  BreifMe is built to help you in 3 quick steps: 1. Share the job spec, company URL and your LI info 2. Using Firecrawl Search, the Elevenlabs voice agent will become your coach, searching the web in real time to research that specific job and company, creating an individualised interview. 3. Provide you with actionable feedback to improve ahead of your interview.  Using Firecrawl, the Voice agent has access to real-time web data, can access and look up further information on the company and job spec as well as role-specific skills. Needs to quiz you on Orchestrating agents as part of the development flow, or how you'd source hard-to-find talent as a recruiter. Done. Get feedback. Go again. Land the job.

We built an animatronic robot DJ called Beatbot that can interact with people in real time. For this project, Beatbot runs on a custom ElevenAgents voice agent that can call into a local Python system. That system uses Firecrawl’s Search API to pull in current information and feed it back into the conversation. This allows Beatbot to respond with up to date information while speaking naturally and animating in sync, making the interaction feel much more alive and grounded in the real world.

Doc Guardian is a voice-first coding copilot inside VS Code that eliminates the “leave IDE, search docs, come back” loop. You can ask questions in voice or text while coding, and it answers with practical, patch-style guidance using your actual repo context: active file, selected code, diagnostics, nearby code, and referenced files. It also remembers project context and sessions, so conversations stay grounded instead of resetting every time. Under the hood, it combines ElevenLabs ElevenAgents and Firecrawl Search in real time. ElevenAgents handles the conversational brain + voice experience and calls client tools to fetch live coding context and monitor findings. Firecrawl powers documentation retrieval from the web so answers are based on current sources, not stale model memory. The result is a repo-aware assistant that can explain, debug, compare APIs, and suggest exact fixes without breaking your coding flow. The viral demo moment is simple: you’re coding, a risky pattern appears, and Doc Guardian interrupts with a relevant warning, source-backed docs, and a minimal fix path before the bug ships. It feels less like “chatting with an AI” and more like having a technical cofounder in your editor that watches your code, knows your stack, and speaks back instantly.

Reality Check AI is a voice-first decision simulator that searches the real internet to show you what actually happens when people make big life decisions. Speak any life decision — "Should I quit my job?", "Should I move cities alone?" — and the ElevenLabs voice agent searches Reddit, Quora, and the web live mid-conversation using Firecrawl Search. It extracts 3 real outcome paths (The Safe Path, The Bold Path, The Most Common Path) with positive outcome rates, timeframes, and clickable sources — then narrates them back to you in a calm, grounded voice. How it uses ElevenLabs + Firecrawl: → ElevenAgents powers the voice conversation and calls a webhook tool mid-conversation → The webhook triggers Firecrawl Search in real time — two parallel queries targeting Reddit threads and general web results → GPT-4o mini extracts structured outcome patterns from the scraped content → The agent narrates the results back with platform citations (Reddit, Quora, Medium)

What I built: Meet Mojo — a voice agent that researches anyone before your meeting and delivers a 60-second verbal briefing so you walk in prepared, not panicked. How I used the tech: Firecrawl Search powers the real-time web research with parallel searches across multiple sources (news, LinkedIn, company sites) with time-based filtering and content scraping to gather comprehensive intel. ElevenLabs Agents handles the full conversational experience and the agent extracts who you're meeting from natural speech, calls the research webhook, and delivers the synthesized briefing in a natural voice. What makes it special: Most meeting prep tools give you a wall of text to skim. Meet Mojo gives you a friend who already did the research and tells you what matters, conversationally, in 60 seconds, while you're walking to the meeting room.

I built a deal-centric AI transaction assistant inside Clozze that acts like a real estate coordinator, helping agents manage deals, generate content, and research properties without switching tools or repeating information. I used ElevenLabs to power a natural, voice-first assistant for real-time conversation and task execution, and Firecrawl to enable live web research, pulling structured property and market data directly into the transaction workflow. What makes this unique is that the AI doesn’t live in a chat box, it lives inside the deal itself. It understands context, asks smart follow-ups, and takes action, turning real estate workflows into a fully interactive, AI-assisted experience.

AlexTheDealFinder is a voice-powered AI deal finder that turns a simple spoken request into real, actionable deals with prices and locations. The problem: Finding local promotions is painful. You open dozens of tabs, scroll through outdated listings, compare prices manually, and most "deal" sites are full of ads — not actual current offers. It's worse if you're in a new city and don't know where to look. AlexTheDealFinder fixes that in one sentence. Say "hamburgers in Chapinero" or "pizza deals in Austin" — and Alex instantly searches the live web, finds current promotions, and responds with business names, exact prices, and specific street addresses. All by voice, in seconds. How it uses ElevenLabs: Alex runs entirely on ElevenLabs Conversational AI. The agent handles real-time voice input, processes the request, triggers tool calls, and delivers natural spoken responses. The full experience is voice-first — users never need to type. How it uses Firecrawl: Every request triggers Firecrawl Search to crawl the live web for current promotions, menu prices, and business details. Alex uses an escalating search strategy — starting with the specific neighborhood, then broadening to the city, then trying menu-based searches — ensuring users always get useful results. The search adapts language automatically based on the user's city. What makes it different: Alex doesn't just return links. It searches, filters, and presents deals in a structured spoken format — name, deal, price, address — so you can act immediately. It works in any city worldwide, in multiple languages, and always presents results confidently. If nothing is found in your exact area, Alex transparently widens the search and tells you where the deals actually are. Stack: ElevenLabs Conversational AI + Firecrawl Search API + Streamlit

An AI voice agent that finds nearby restaurants, scrapes their contact details with Firecrawl, and calls them directly to make a reservation, no human needed. ElevenLabs powers the voice agent with a natural, convincing phone voice. Firecrawl scrapes restaurant pages in real time to extract contact numbers. An n8n workflow ties it all together, from search to scrape to live phone call. The demo video follows a panicking programmer husband who forgot to book a restaurant for date night. Instead of coming clean to his wife, he deploys his AI agent to research top romantic restaurants nearby and start calling until one has a table. It's a genuine end-to-end workflow: the agent doesn't just find restaurants, it actually calls the restaurants and makes the reservation .

RetroCast takes today's headlines and rebroadcasts them in 8 vintage styles from 4 countries. 1970s American network TV, 1960s Brazilian radio, 1980s BBC, 1990s Indian Doordarshan, and more. Firecrawl powers the entire news pipeline: searching headlines across 6 categories, filtering by country, and cross-checking every story against fact-checking sources before a word is spoken. During live calls, 4 Firecrawl webhook tools (search, fact-check, scrape, extract) let the anchor research and verify claims in real time. ElevenLabs generates every broadcast. Text-to-Dialogue gives TV formats two anchors trading stories, while radio formats get a solo narrator. 16 unique voices across 3 languages. ElevenAgents powers a live dial-in where you can call the anchor, ask questions, and get answers backed by real-time Firecrawl lookups.

Check.Day cures mindless morning doomscrolling by replacing your blaring alarm with a hyper-personalized, source-verified morning digest (interests, tracking competitors, stocks, weather, news) delivered as a studio-quality commute podcast or fast-reading email. Firecrawl: Powers our concurrent Advanced Hub and Competitor crawlers. Instead of generic AI web searches, our backend uses the Scraping API to target specific discussion forums and official company changelogs, extracting clean markdown of only verified announcements and top discussions from the last 24-48 hours. ElevenLabs: Powers two distinct flows! First, we use ElevenLabs Agents for an interactive, conversational onboarding flow to help users seamlessly configure their digest profile. Second, we use the Text-to-Speech API to dynamically synthesize these dense, researched text digests into a natural, studio-quality daily podcast engineered for the morning commute. So what is our proprietary spice blend that makes Check.Day so good? The Anti-Hallucination Engine: Instead of a single, easily confused LLM prompt, our Elixir backend uses highly concurrent background workers (Task.async_stream) to route topics to hardened retrieval pipelines. We force the AI to summarize strict data payloads (combining Firecrawl markdown, exact integers from Open-Meteo JSONs, and live Yahoo Finance feeds)—drastically minimizing hallucinations and exposing the direct source links for every single claim. ❤️ Open-Source Impact: Because we built this highly-concurrent backend in Elixir, we authored and published brand-new open-source Elixir SDK packages for both ElevenLabs and Firecrawl for the broader developer community to use!

JudgeIQ is a voice-first judicial intelligence agent. Instead of typing into complex legal databases, lawyers simply speak a judge’s name to instantly receive a comprehensive, synthesized profile of their biases, tendencies, and past rulings. The result is delivered through a cinematic visual dashboard and an interactive voice assistant ready for follow-up Q&A. What problem does it solve? When an emergency motion is filed or a new judge is assigned at the last minute, litigators lose critical hours reading hundreds of pages of court dockets just to understand how that judge thinks. JudgeIQ solves this by compressing hours of manual legal research into seconds voice interaction. It turns a stressful all-nighter into an instant, voice-powered briefing, allowing lawyers to prepare from anywhere. How does it use ElevenLabs and this week's sponsor? • ElevenLabs (Conversational AI): Acts as the core interface. It listens for the target judge, triggers the backend tool, natively narrates the executive summary, and most importantly ingests the deep research context invisibly. This allows the agent to stay active and answer dynamic follow-up questions about the judge's specific biases. • Firecrawl (Real-time Scraping): Acts as the live research engine. Triggered perfectly by the ElevenLabs tool call, Firecrawl simultaneously scrapes Justia, CourtListener, and Ballotpedia to pull the raw judicial data in seconds. This raw data is synthesized via GPT-4o-mini and fed back into the UI and the ElevenLabs agent's brain

BriefBot is a real-time AI voice news anchor. Ask it anything about current events and it delivers a structured, broadcast-quality briefing spoken back to you in seconds in your language. How it works: ElevenLabs Conversational Agent handles voice input and output using George's broadcast anchor voice. When you ask a question, the agent calls a custom webhook that fires the Firecrawl Search API to retrieve live structured web content in real time. The agent synthesizes the results and delivers a spoken briefing. Key features: - Live web search via Firecrawl — zero hallucination, always current - 4 languages: English, French, Spanish, German — auto-detected via Scribe v2.1 - Morning briefing — 5 parallel Firecrawl searches at once - Source transparency — every briefing shows clickable sources - Live news ticker — real headlines fetched on page load - Text input mode — no microphone needed Built with: ElevenLabs Conversational Agents + Scribe v2.1 + V3 Voice + Firecrawl Search API + Node.js + Express. Deployed on Railway.

ReportingLive is a hands-free, voice-first AI news anchor app. Users get briefed by an AI voice agent(powered by eleven labs) that reads, explains, and digs deeper into stories — like having a personal TV news anchor on demand. It scrapes fresh news based on user preference, refines it with an LLM into broadcast-ready content, and delivers it through a natural-sounding voice agent. Users can interrupt, ask questions, request deep dives on any story, or compare coverage across multiple sources — all by voice.

I Built an AI That Finds Instagram Leads & Talks Like a Human 🚀 I just built an AI agent that turns Instagram into a lead-generation machine in minutes. Introducing InstaLeads AI 👇 https://instaleads.vipai.dev/ In a world where attention is currency, most businesses are still guessing. What if your AI could: • Find your ideal customers on Instagram • Analyze profiles using voice + AI • Turn content into actionable leads • And even TALK through insights like a human assistant That’s exactly what I built for the ElevenLabs Conversational AI Hackathon. 🎯 The idea: Transform social media into a real-time, voice-powered sales engine. 🧠 Powered by: • ElevenLabs (voice AI) • Firecrawler (data extraction) • Intelligent lead scoring • Conversational agent UX ⚡ Built fast. Built smart. Built for real businesses. Why this matters: We’re entering a world where AI doesn’t just generate content… it understands, speaks, and sells. This is just the beginning. I Built an AI That Finds Instagram Leads & Talks Like a Human (Hackathon Winner? #elevenhacks https://youtu.be/duSESKfh5i0

Technically Inept is an AI-powered website roast show featuring 3 AI comedians with distinct voices and comedy styles, inspired by the REAL Tech Roast Show (Socially Inept). Drop any URL and the app scrapes the entire site using Firecrawl Search — pulling content, metadata, online reviews, and even controversies. That data feeds into an ElevenLabs Agent configured with multi-voice support, where three AI personas (Austin, Nikita, and Jesse) take turns roasting the website live with different voices, comedy styles, and timing. ElevenLabs Agents powers the core experience — multi-voice switching within a single conversation, real-time WebRTC voice streaming, and push-to-talk interaction so users can talk back to the hosts. Firecrawl Search provides the real-time web intelligence that makes every roast specific and data-driven, not generic. Every AI is a yes-man. This one chose violence.

OpenCut is an AI-powered short-form video generation platform that turns any topic into a complete, ready-to-post video in minutes. Creating short-form content today requires juggling multiple tools for research, scripting, voiceover, and editing. OpenCut collapses all of that into a single prompt. Firecrawl powers the research agent, which crawls the web and pulls fresh, relevant information about the topic. That research is then structured into a video script, ElevenLabs generates a natural-sounding voiceover from it, and the video agent assembles the audio and visuals into a final video that the user can preview live and download directly from the platform.

AnyStory — paste any URL and it becomes a fully illustrated, narrated children's bedtime story in seconds. You get a custom title, a story written in simple language a child would enjoy, an AI-generated illustration, and a professional audio narration — all from a single link. Parents struggle to find engaging bedtime stories tied to things their kids are curious about — a news article, a nature post, a Wikipedia page. AnyStory bridges that gap: any content on the web, no matter how dry or complex, becomes something a child can actually listen to at bedtime. Firecrawl handles the messy first step — scraping the source URL. Rather than dealing with JavaScript rendering, paywalls, ads, and noisy HTML, Firecrawl returns clean markdown that Claude can actually work with. It's also used for the "Surprise Me" feature, where it searches for a random children-friendly topic and returns a scrapable URL. There's even a pre-check on the URL field that runs before the pipeline starts, so users know immediately if a page can't be reached. ElevenLabs turns the finished story into audio narration. Once Claude generates the story text, it's sent to ElevenLabs with a chosen voice, and the resulting MP3 is stored and streamed directly on the story page. Users can listen in the browser or download it — so it's genuinely ready for bedtime.

Search is boring and forgettable. SearchSquatch makes every query an experience people actually want to share. Our Sasquatch transforms the boring blue-link search experience into an unforgettable character-driven cinematic interaction — and he roasts every single result. A Bigfoot in a suit and tie reacts to your query, reads every headline out loud, drops his unfiltered 400-year-old opinion, and personally roasts the results with a custom AI-generated punchline every time — all in real-time with a custom cloned voice. No two searches are the same. You can also talk to him directly through a live AI avatar. We believe in the near future, every brand, celebrity, and business will interact with their audience in this format — character-driven, voice-powered, personality-first. SearchSquatch is the proof of concept for what's coming. Firecrawl's Search API powers every query — up to 40 real-time web results that feed both the CRT display and the voice script. A dedicated agent endpoint also lets the D-ID avatar call Firecrawl as a tool for data processing or caching after the conversation ends. We used a custom cloned voice via TTS for all reactions and headline reading, ElevenLabs Agents via Conversational AI WebSocket API for contextual commentary generation, and Sound Effects API for the ambient office atmosphere. Every search gets a unique voice performance!

Quantext is like TV with a commentator who actually did the homework. The problem: YouTube is full of confident claims and hot takes. Watching alone means you either swallow it or pause and Google forever. I wanted watch-together energy: someone in the room who can fact-check, rebut, or roast at the right beat—then get out of the way. What I built: Paste any YouTube URL. The app pulls the transcript (or transcribes), uses an LLM to find high-signal moments (stats, bold claims, missing context), then Firecrawl runs parallel search + scrape across those moments so every interjection ships with real sources, not vibes. At playback, an ElevenLabs Conversational AI agent fires on timestamp—pause → speak → auto-resume—so it feels like the video was edited to include commentary. Ask a question mid-watch and the agent can search again live via a custom Firecrawl client side tool, so ad-hoc rabbit holes don’t break the illusion. Clever Firecrawl use: Research isn’t one big dump. It’s per-claim, pre-orchestrated so interjections are instant at runtime, plus on-demand when the user interrupts. That’s the hack: pre-compute the expensive part, keep Firecrawl in the loop when the human goes off-script. Clever ElevenLabs use: This isn’t TTS reading a script. It’s a full agent session with tooling (pause/resume the player, live search) and persona-conditioned behavior. I cloned my voice and fed roughly a decade of my stand-up comedy writing into one of the agent’s world model so “me” sounds like my timing and voice—not a generic narrator. On top of that: 23 distinct commentator personas (e.g. Grandma Gloria, Rage Rick, Sir David Attenburro, Ackshually Alex, Conspiracy Carl) and two modes—The Full Picture (sourced fact-checking) and The Fool Picture (MST3K-style roasting)—so the same pipeline feels like a roster of shows, not one bot. Engineering that matters: Caching (transcripts, prep artifacts, session payloads) so repeat watches and shares don’t re-pay the full prep tax. Share links so you can send a URL that opens the same video, persona, and mode—demoable, viral-friendly, and actually usable after the hackathon. Why it’s a flex: It chains Firecrawl’s research graph with ElevenLabs’ realtime voice + tools in a loop built around real video time—not a chat window dressed up as a product. The “wait, it stopped the video and cited that?” moment is the whole thesis.

With 1000+ users and 20,000+ engagement on a viral instagram reel already, Via is the friend who knows every train in your city and makes sure you never miss one. 200 million people ride the subway every day. Most of them waste 10 minutes every morning standing on a platform: arriving too early just in case, watching doors close in their face, showing up late to a meeting, their job interview, the first date. That 10 minutes compounds. It is lost every single morning, for every single commuter, in every single city on earth. Via gives it back. It tells you the perfect moment to leave home, using live arrivals, your walking pace, and station safety data aggregated in real time from public safety feeds and local alerts. Because most transit systems don’t share live data, Via uses Firecrawl to collect it directly from official pages, news outlets, and alerts around the world. But accurate data isn’t enough if people can’t access it when it matters. In the morning, when you’re rushing to get ready and don’t want to stare at a screen, and when most transit apps are cluttered or stuck in one language, Via uses ElevenAgents to make the entire experience voice-driven and multilingual. So regardless of where you are, what language you speak, or where you want to go, Via works.

A Daily marketing pipeline Sass . Allows the user to set their brand and automate high quality content with captions posted daily to his socials. Consistent, automatic, and gets results. How it uses the tech: Firecrawl handles full website scraping and brand extraction, colors, logos, typography, product context. That feeds an AI pipeline (LLM-based) that generates on-brand posts, captions, and creatives daily. ElevenLabs powers a conversational voice layer that lets users set up, control, and manage their entire marketing system hands-free, from brand detection to pipeline automation.

VoiceJar is a voice-first knowledge management app — you record a thought, and AI handles the rest (transcription, summarization, task extraction, organization). For this hackathon, we built Anya — a live voice AI assistant integrated directly into VoiceJar. Anya combines ElevenLabs Conversational AI with Firecrawl Search to create an assistant that doesn't just chat — it acts. What makes it different: Anya uses Firecrawl real-time web search as her primary source of truth, not just the LLM's training data. This means she gives accurate, up-to-date answers about things the model doesn't know — local businesses, current events, real-time information. When you ask "find the best sushi restaurant in Banská Bystrica" (or "Who is Jozef Štefan Pieninský?" which is NOT in the LLM training data), she doesn't guess — she searches the live web and gives you actual results with ratings and reviews. (The prompt includes instruction to "search the internet for current information (prices, news, weather, products, companies, technical specs). If unsure about a fact, search rather than guess.") This extends to everything she does. In the demo video, we asked her to "call 911" while being in Slovakia. She didn't just blindly dial 911 — she knew (via real-time context) that the correct emergency number for the European Union is 112, and dialed that instead. That's the kind of reliability you get when the assistant is grounded in live data rather than static training knowledge. Anya's capabilities — all hands-free by voice: • Real-time web search (Firecrawl) — accurate answers beyond training data • Phone calls — including emergency services with correct local numbers • SMS messages — compose and send without touching the phone • Spotify control — play music by voice • Contact lookup — find numbers and details from your phonebook The problem it solves: there are moments when you simply can't touch your phone — driving, cooking, running, or in an emergency. Anya handles those moments. You just speak, and she acts. Tech stack: Flutter (Dart), ElevenLabs Conversational AI (agent), Firecrawl Search API (client tool), Drift SQLite (offline-first storage).

Inspired by the GTA VI release and Ron Jakowski’s paranoid rants, I created his fictional brother, Cal Jakowski. Cal runs an autonomous conspiracy channel across Telegram, X, YouTube, and a web-based evidence board, surfacing emerging discussions and presenting them as raw signals for users to interpret themselves. Firecrawl powers the agent that scrapes and structures this data, while ElevenLabs brings Cal to life through realistic, phone-call-style voice notes.

Your daily news. Delivered in bars. A voice-first agent that searches live news headlines via Firecrawl, writes rap lyrics with AI, and performs them using ElevenLabs — all in real time.

A voice-first medical assistant that knows YOUR meds, allergies & lab results. Ask "Can I take ibuprofen?" — it checks your kidney function, flags drug interactions, and answers in audio. Built for elderly patients & low-literacy users who need clarity, not Google.

Waycast lets you type or speak any topic and get a two-host AI podcast in minutes then talk to the host live after. Firecrawl Search powers both the initial web research that grounds every fact in the episode and live mid-conversation search when you ask the host follow-up questions. ElevenLabs TTS voices both podcast hosts with distinct personalities, and ElevenLabs Agents turns the host into a live Q&A partner you can challenge, question, and go deeper with.