Skip to content

docs(gtm): record the model-to-production thesis, checked against the repo - #187

Open
unfoundbox wants to merge 1 commit into
mainfrom
claude/sweet-meitner-wqn64h
Open

unfoundbox wants to merge 1 commit into
mainfrom
claude/sweet-meitner-wqn64h

Conversation

@unfoundbox

@unfoundbox unfoundbox commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

What

  • docs/gtm/model-to-production-thesis.md (new): records the 2026-10-02 session summary. One-line description: autonomous model-to-production infrastructure. Business model: open core plus inference engineering as a service. It covers the architecture loop, the Qwen / p95 < 300 ms wedge, the GTM, and the plan to turn job descriptions into evals.
  • It sets this beside the same-day "local lane for your coding agent" recommendation. Both use the same core but aim at different buyers.
  • It maps each thesis layer to what exists on main at 9bc8a2d. The registry, receipts, MCP server and a sweep harness seed (sweep.py) exist. The registry has no vLLM, SGLang or TensorRT-LLM entries and no datacenter-GPU measurements. The planner, docks and providers are zero code.
  • It names the hardware ceiling. The quota allows one g6e.2xlarge, so one L40S. An H100 comparison needs a quota request or a box rented elsewhere.
  • docs/DECISION-INBOX.md: a new entry asks which buyer leads, the developer's local lane or the team that needs inference owned.

What this does not do

It leaves VISION.md, the README and the landing page unchanged. They wait on the decision above.

🤖 Generated with Claude Code

https://claude.ai/code/session_011TqzXL1YTm2Dk55TQyk1JS


Generated by Claude Code


View with [code]smith Autofix with [code]smith
Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.

… repo

Captures the 2026-10-02 session summary (autonomous model-to-production
infrastructure, open core plus inference engineering as a service), sets it
beside the same-day local-lane recommendation, and maps each thesis layer to
what exists on main. Logs the lead-buyer question in the decision inbox;
VISION.md is left unchanged until that is decided.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011TqzXL1YTm2Dk55TQyk1JS
@vercel

vercel Bot commented Oct 2, 2026 •

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated
spacepilot Error Error Oct 2, 2026 10:03am UTC
spacepilot.dev Error Error Oct 2, 2026 10:03am UTC

Copy link
Copy Markdown
Contributor Author

The red Vercel – spacepilot status isn't from this PR. The PR only adds two Markdown files under docs/. The same status failed on the head of #183, which merged on 2026-09-25 (25e7c4e). The Vercel spacepilot project has no rootDirectory set (the landing site deploys from landing/ through the spacepilot.dev project), so its failure looks like a project-settings problem, not a code one. I haven't read the deploy logs, so that isn't confirmed. Nothing in the repo fixes it yet. The likely fix is in Vercel: delete or disconnect that project, or point it at a real root. The GitHub Actions workflows (pytest, packaging, security) didn't run on this PR because its diff only touches docs.


Generated by Claude Code

This branch had an error being deployed

2 failed deployments
Preview – spacepilot.dev — 7694b6cb Deployed Oct 2, 2026 by vercel[bot]
Preview – spacepilot — 7694b6cb Deployed Oct 2, 2026 by vercel[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants