Bhavya

TechSnark is a voice-powered AI agent that solves the critical problem of AI assistants having outdated training data by combining ElevenLabs conversational agents with real-time web search via Firecrawl API and Claude 3.5 Sonnet intelligence and replit .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.

Submitted 22 Mar 2026