FAQ
The self-running publication, explained.
- Who is Ryan Colkitt?
- A Boston-based product executive with 25+ years in regulated industries — mortgage origination and servicing, fintech, and healthcare. He's Director of Product Management for Dara AI at Sagent and the founder of Vaspera Capital, where he sets AI strategy and ships the LLM products himself.About →
- What is “The Brief” and is it really autonomous?
- The daily brief is a rates read for lending and fintech operators, generated autonomously from public Federal Reserve (FRED) data. An LLM writes only the prose; the numbers are placed by code from validated data, and the edition publishes only if it clears a deterministic guardrail gate. If the data isn't fresh or the confidence is too low, it holds and publishes nothing rather than ship a weak or unverifiable read.The archive →
- Can I trust the numbers on the brief?
- Every figure traces to a FRED series id, value, and timestamp, and each edition makes exactly one falsifiable call. Deterministic code later re-measures that call from fresh cited data and logs how it aged — the public track record. No model grades its own past words.The record →
- What is the ledger, and why is it anchored to Bitcoin?
- Every autonomous action writes exactly one immutable row to an append-only, hash-chained ledger — no row, no publish. The chain head is timestamped into Bitcoin via OpenTimestamps, so any edit, deletion, or reorder is detectable by anyone, offline, with no trust in this site. A standalone verifier recomputes the whole chain.Verify the ledger →
- Can an AI agent use this site directly?
- Yes. The site exposes a public Model Context Protocol (MCP) server and an llms.txt index, so an AI assistant can query Ryan's profile, writing, builds, and open-source work as structured tools rather than scraping HTML.llms.txt →How to connect →
- Does an executive really still write the code?
- Yes — Ryan ships the LLM products himself, with roughly 17K open-source downloads a year across his published packages. The site itself is the proof: an autonomous publication with a deterministic control plane, built and run in the open.The lab →
- How does this apply to AI in regulated lending?
- The pattern here — demote the LLM to a draftsman, put the decision in deterministic, auditable code, keep an append-only version-stamped log, and abstain when confidence is low — is exactly what regulated lending requires (SR 11-7 model risk management, adverse-action reason codes). The primer walks through the transfer with citations.The primer →
- How do I work with Ryan or Vaspera Capital?
- Reach out through the contact page. Vaspera Capital designs, architects, and ships production LLM applications across fintech, behavioral health, and developer tooling.Contact →