softwarefactory.build

A directory and guide library for agent-native software factories.

Directory / Trevize

Trevize

Verified

Shared cloud workspaces where teams and agents close the loop.

Trevize runs coding agents in isolated cloud workspaces that a product team can join live. Each session deploys the repositories and preview resources the change needs, then keeps iterating on failed checks until a pull request is ready. Humans still merge. The product is built for shared work, not a single developer chatting with an agent on a laptop.

Model independence
Multi-model
Deployment
Cloud
Open source
No
License
Proprietary
Pricing model
Mixed
Added / updated
19 Aug 2026 / 19 Aug 2026

trevize.dev

Editorial take

Trevize is one of the few products that starts from the factory problem — a shared computer, a preview, a loop that does not stop at a plausible diff — instead of bolting cloud agents onto an IDE. The public site is unusually specific about isolation, authenticated previews, recovery from provider limits, and the rule that a human merges every pull request. It is also early. There is no public source, no independent case study, and no published production metrics. This directory's publisher founded Trevize; read the listing as an editorial description of a product we operate, not as a third-party audit. See Conflicts of interest on the methodology page.

Strengths

  • Isolated cloud workspaces with authenticated team access and live previews, rather than a hidden local agent session.
  • Blueprints that pin prompt, harness, model, and pass conditions per step, then loop on failure.
  • Bring-your-own inference via API keys or Claude Max, OpenAI, and Cursor Max subscriptions; Trevize credits are billed for compute.
  • Signal connectors the marketing site documents for Linear, Slack, and Sentry, plus Neon or Supabase database branches per session.

Limitations

  • Cloud-only. No self-hosted or air-gapped deployment is documented.
  • No public repository or published license text beyond a proprietary product.
  • Starter plan caps credits and concurrent runs; Team and Scale add usage overage at $0.04 per credit, per the pricing table on 19 August 2026.
  • Production evidence is absent. "Used by product teams at" on the homepage is marketing copy, not a verifiable case study.

Compare

  • vs factory — Two products that sell the factory loop, not an IDE sidecar.

Fields last checked against primary sources on . Pricing and plan names rot; follow the vendor URL.