TypeSafe AI ships cookbooks: batching questions cuts cost 12.2x and runs 10x faster
hackgoofer · x · 2026-10-03
TypeSafe AI released hands-on cookbooks showing its confidence-based Q&A primitives in action: self-consistency routing uncertain outcomes to human review; batching 13 questions into one call on a GDPR briefing for 12.2x cheaper and 10.0x faster execution with unchanged answers; and BM25 re-ranking lifting CLERC legal query top-1 accuracy from 5% to 18% and top-10 from 38% to 62%. ZDR and higher rate limits are available via sales.
More from coding & agent
- VoidZero founders on TypeScript.fm: type-aware linting goes stable, Cloudflare deal and Vite's future — cnakazawa · 2026-10-03
- The default AI coding stack: OpenAI+Anthropic models on Stripe, Supabase, Vercel infra — alexmacgregor__ · 2026-10-03
- Reddit MCP Server gives AI assistants key-free access to Reddit search, posts and user data — modelcontextprotocol · 2026-10-03
- Frenchie MCP connector adds OCR, transcription, file extraction and image gen to agents — modelcontextprotocol · 2026-10-03
- Stripe's upcoming agentic identity API may let trusted agents claim free trials for users — jeff_weinstein · 2026-10-03
- repligate demos memory system spanning 72.9M tokens, letting a model retrieve raw past experiences — repligate · 2026-10-03