# TenRail > TenRail is the 10x delivery method for teams that code with AI agents. AI made coding fast; TenRail makes shipping fast. Code is about 20% of a task, so AI tools alone give about 1.2x. TenRail removes the other 80% (finding the rule, waiting on decisions, waiting on other teams, rework) and has teams measure their real factor in their own repo over two sprints. 10x is the planning target, never a guarantee: below 4x the plan is regenerated with the measured number. ## Offer - TenRail Individual: one-time payment of $129 USD with lifetime updates; founder price $99 USD until December 25, 2026. Team plan (up to 10 people): $389 USD, founder price $299 USD. Company plan: on request. 7-day money-back guarantee. Includes the method (18 chapters in English, Portuguese and Spanish), 22 templates, 21 Claude Code skills, 12 agents and the tenrail CLI. - 7-day money-back guarantee, no questions. - Team: licences for a team, one invoice (contact talk@tenrail.dev). - Adoption: guided rollout for platforms and legacy replacement, priced per engagement (contact talk@tenrail.dev). - Works with Claude Code, Cursor, Codex and Gemini (any agent that reads Markdown instructions). ## How it works - Spec before code: approved specs with business rules and acceptance criteria in Given/When/Then; every test cites the criterion it proves. - Decisions before the task: a decision log with recommendation and priority; tasks tied to open blocking decisions don't start (tenrail dor). - Contract and fake first: each bounded context is a lane whose first task publishes interfaces and an in-memory fake, so people and AI agents build in parallel. - Gates instead of opinions: tests per criterion, mutation testing, architecture fitness functions, agent-run E2E, same-day human review (tenrail trace). - Measure instead of promise: factor = baseline effort recorded before the task ÷ effort spent, only for tasks with a complete Definition of Done (tenrail measure). ## The 10x activity model Reference model: 10% of the original effort stays in the task, so the model factor is 10.0x. With AI tools only (coding and tests faster, everything else unchanged) the same model gives 1.2x. Drop the approved spec and it falls to 3.0x. These are starting assumptions; each team replaces them with its measured numbers. ## Adoption levels - N1 Solo, N2 Product team, N3 Platform and software houses, N4 Legacy replacement. ## Tools - tenrail CLI (Python 3.11+, one file, no dependencies, runs in CI): init, lint, dor, trace, roadmap, measure, model, check. Reads specs in English, Portuguese and Spanish. - Installed in two commands: ./install.sh "", then python3 tools/tenrail.py check. ## Origin Built while planning the replacement of two legacy platforms: ~1,000 tables, ~180 scheduled jobs. ## Pages - [Home (English)](https://tenrail.dev/) - [Português](https://tenrail.dev/pt/) - [Español](https://tenrail.dev/es/) - [Pricing](https://tenrail.dev/#pricing) - [FAQ](https://tenrail.dev/#faq) ## Links - GitHub: https://github.com/TenRail - Instagram: https://www.instagram.com/tenrail.dev/ - Contact: talk@tenrail.dev