Trying the Software Factory Pattern
Imprint's engineering lead details adopting an AI 'software factory' pattern that loops agents on Linear projects using Datadog metrics.
The post recounts Imprint's 2026 adoption of AI coding agents: company-wide Claude Code usage, roughly 10 local workspaces for cross-repository pull requests, a migration from Jira to Linear, and an orchestrated 'Agent Fleet' harness modeled on Stripe's Minions. The software factory pattern has an agent audit a Linear project's goals via Notion RFCs and Datadog/Snowflake metrics, then create, update, and complete issues autonomously. The author also runs the loop in a less frequent post-release mode to catch adoption or error-rate changes on shipped projects like a passkeys implementation.
- Imprint rolled out Claude Code company-wide, then moved ticketing from Jira to Linear for agent visibility.
- An 'Agent Fleet' orchestration harness handles tasks, following Stripe's Minions model.
- Factory loop audits goals via Notion RFCs and Datadog or Snowflake metrics before working tasks.
- Post-release mode monitors shipped projects such as a passkeys implementation.
- Products
- Claude Code
- Organizations
- Imprint
From the original article
3 min read at lethain.com
One of the interesting challenges of the AI ecosystem in 2026 is that new, effective patterns emerge faster than I can adopt them. I’ll find a handful, get back to work, and realize a month later that I’d missed four or five more.
By gpi · lethain.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.