Urgent.News

What's breaking now, across thousands of outlets.

AI

I Built My Friend a Voice-First AI for Unfinished Thoughts

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend What I Built I built Thread for my friend, who often gets ideas while walking, reading or watching a series. One afternoon, an idea surfaced while he was already concentrating on something else. He decided to write it down after finishing rather than interrupt the work in front of him. When he finally opened his…

This submission is for the Hacktoberfest Weekend Challenge: Build for a Friend What I Built. The author created Thread, a voice-first AI thought companion for capturing ideas that surface before they are fully formed. The user speaks naturally, and Thread records the audio, transcribes it with ElevenLabs Scribe, and stores both the original transcript and AI-generated summary, categories, possible action, and question.

Thread then checks for other related thoughts by converting transcripts into embeddings and searching the posted timeline, using pgvector to find up to five relevant earlier thoughts. A Gemma model then examines whether these relationships could be helpful. The AI does not decide what the user intended but offers possible connections, implications, and questions that could help the user explore those ideas further. The original words remain visible along with the AI interpretation.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Research Reports Your Agent Writes Need Sources Before Sentences

Ask a coding agent for a research report and you usually get fluent prose first and evidence later — if at all. A single forum post becomes "research shows." A number from 2023 is presented as the…

  • Research reports require vetted sources before finalizing sentences
  • Workflow mandates defining report question and sub-questions first
  • Tertiary sources like social media cannot support core conclusions

I gave my text-to-SQL agent a business glossary. One version helped a lot, one did nothing.

A text-to-SQL agent sees table and column names. It does not know that "revenue" in your company means billing_invoice.total_net , only for invoices with status = 'issued' .

  • Text-to-SQL agent improved by 9.9 percentage points with complete business glossary
  • Glossary enabled model to pull relevant tables, despite ineffective learned terms from other users
  • Focus on documenting frequently debated business terms like revenue and active customer

More from Monday 5 October →