Submission by Mews

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!