# Stewart's Notes

A quiet personal blog about reliable AI systems, accessibility, and practical software work.

This is a plain-text index for AI indexing/discovery. Canonical post sources live as Markdown or MDX files in git under src/content/posts/.

## Posts

- [Making AI Reliability Accessible](https://discostew-blog.pages.dev/posts/ai-reliability-accessibility/)
  - Published: 2026-06-09
  - Summary: A short note from Stewart on treating AI reliability as an accessibility practice, not just an infrastructure concern.
  - Tags: AI reliability, accessibility, software practice

- [Fidelity over backward compatibility](https://discostew-blog.pages.dev/posts/fidelity-over-backward-compatibility/)
  - Published: 2026-06-09
  - Summary: Why I accepted breaking API changes in Propontis when the old schema hid important evidence boundaries.
  - Tags: Propontis, API design, auditability, red teaming, AI reliability

- [I fixed an eval by hashing the contract, not the file](https://discostew-blog.pages.dev/posts/hashing-the-contract-not-the-file/)
  - Published: 2026-06-09
  - Summary: Why an evaluation baseline should track the public behavior users depend on instead of private implementation bytes.
  - Tags: evals, AI reliability, testing, provenance

- [Let the LLM write the bio. Make the database check the facts.](https://discostew-blog.pages.dev/posts/let-the-llm-write-make-the-database-check-facts/)
  - Published: 2026-06-09
  - Summary: A design note on generated biographies, stat-claim extraction, and using structured data to keep prose honest.
  - Tags: LLMs, verification, baseball RAG, AI reliability

- [MCP servers have a context tax](https://discostew-blog.pages.dev/posts/mcp-servers-and-context-tax/)
  - Published: 2026-06-09
  - Summary: Why I do not want persistent MCP servers in my default Hermes profile unless the recurring value is obvious.
  - Tags: Hermes, MCP, local AI, agent workflows, stewardship

- [The model should explain the evidence, not become the evidence](https://discostew-blog.pages.dev/posts/model-should-explain-evidence-not-be-evidence/)
  - Published: 2026-06-09
  - Summary: A short decision story about source authority, provenance, and keeping generated prose out of the truth path.
  - Tags: AI reliability, provenance, RAG, source authority

- [Removing ChromaDB made my RAG system more reliable](https://discostew-blog.pages.dev/posts/removing-chromadb-made-rag-reliable/)
  - Published: 2026-06-09
  - Summary: Why I removed a vector database from my baseball RAG project and kept DuckDB as the source of truth for structured facts.
  - Tags: AI reliability, RAG, provenance, DuckDB, accessibility