Tanay Venkata

Home

Builds

  1. Stemma

    Frontier models know RAG techniques, but that isn't enough for production systems in high-risk fields like law, regulation, finance, and healthcare.

    Started ~Dec 2025 · in progress · build

  2. FAA Directive Impact

    FAA orders name engine models, but only some of those engines need the work. Sorting out which ones, starting with a single engine family.

    Started Sep 2026 · in progress · build

  3. Mise

    Keeps the pantry, spice drawer, and equipment on file for ChatGPT, down to the half jar of tahini, so dinner gets planned from what is actually there.

    Sep 2026 · build

  4. FDA Guidance Diff

    Lines up two versions of an FDA guidance, sometimes eighteen years apart, and sorts each change as tighter, looser, clearer, or new.

    Jan 2026 · build

  5. Campaign Intel

    Searches thirty-seven focus-group transcripts from two election cycles by what was meant, not the words used, grouped by who was in the room.

    Jan 2026 · build

  6. SEC Risk Analyzer

    The part of an annual report where companies list what could go wrong. Search five years of it from five large companies and see what quietly changed.

    Dec 2025 · build

  7. 401(k) Plan Advisor

    A reader for 401(k) paperwork that knows 'can I buy stocks' and 'self-directed brokerage window' are the same question.

    Dec 2025 · build

  8. Liar's Dice Engine

    Liar's Dice, it turns out, needs no AI, only a player who knows the odds. This one beat 77 of 80 opponent setups across 32.4 million rounds.

    Nov 2025 · build

  9. AI Study Group

    Three AI agents with different temperaments work through a topic by leaving notes for one another, and stop when they agree.

    Apr 2025 · build

  10. Concept-to-Page

    Describe a product and a landing page comes back in about three minutes, researched enough to avoid the generic and rough enough to argue with.

    Apr 2025 · build