AI Engineering Intensive · Pragma Works
Not a course. Not slides. Two days on your actual codebase — using a published, DOI-backed methodology tested across six production projects.
Book a discovery call30-min call · No pressure · We'll tell you if it's not a fit
⬡ 3 slots remaining in Q2 2026
Proof of work
Six production cases. One controlled replication study. All results traceable to committed repository history — not a demo environment, not a toy project.
The problem
There's a measurable difference between "we use Copilot" and "we ship production-grade AI-assisted code." Most teams plateau at autocomplete. The Forge closes that gap in two days.
The workshop
Day 1 gives your team the muscle memory. Day 2 applies it to the code they'll push to production next week. The same methodology runs both days — spec-driven development, SOLID as AI navigation, TDD as safety net.
Teams build a working, spec-driven prototype from a blank directory. Every participant ships something deployed by end of day. The "how did we build this in 8 hours" moment happens before dinner. By 4pm, teams that have never shipped AI-assisted code are building real applications from specification alone.
Day 2 applies the methodology to code your team owns. Automated test generation, documentation extraction, SOLID refactoring with AI, architectural quality scoring. Teams don't practice on a demo — they improve a real system they'll push to production next week. Rescue, migration, and takeover scenarios handled here.
One week before, we assess your codebase architecture, test coverage, and AI readiness. You know exactly what the two days will tackle before anyone books a flight.
A written methodology document tailored to your stack and processes. Not generic advice — the specific patterns your team follows going forward.
ForgeCraft and CodeSeeker installed and configured for your environment before Day 1. Your conventions, your workflow — not workshop utilities that stop working Monday morning.
A dedicated Slack or Teams channel. Questions answered within one business day. Adoption doesn't stop when the workshop ends.
The methodology
A repeatable quality loop — not a one-time setup. Every session follows the same five steps. The loop is enforced by tooling, not discipline.
| Step | What happens | What it prevents |
|---|---|---|
| 01 Project Setup | AI reads your spec, detects project type, generates engineering standards and ADR-000. One command. | Architecture drift from session one |
| 02 Readiness Gate | Five checks pass before any code is written: spec · architecture · constitution · ADRs · use cases. | Building on an unstable foundation |
| 03 Implement | TDD loop with a bound session prompt. RED → GREEN → REFACTOR. Documentation cascade on every passing commit. | Tests written after the fact, if at all |
| 04 Quality Audit | Score 0–100. Surfaces file bloat, hardcoded secrets, layer violations, and missing specs before code review. | Problems discovered in production |
| 05 Close Cycle | Re-validates all gates, records the convergence score, promotes reusable checks to the community registry. Repeats. | Quality degrading sprint over sprint |
Generative Specification — the published, DOI-backed methodology behind The Forge. Seven formal properties. A quality score out of 14. Six production case studies with verifiable commit history.
Read the white paper →Who it's for
Series A–C startups with 10–50 engineers who need to compress months of AI adoption into two days. CTO is on the call. Budget is pre-approved.
Mid-market SaaS teams where competitors are shipping AI-assisted features faster. Board is asking questions. VP Eng needs a credible answer by next quarter.
Teams who inherited a codebase nobody understands. Contractor delivered a zip file. Lead engineer left. Production system, no documentation. We have a specific methodology for this.
The offer
LATAM engagements — pricing discussed on discovery call.
⬡ 3 slots remaining in Q2 2026
Book a 30-min discovery callDay 1 guarantee: if your team doesn't have a working prototype deployed by end of Day 1, you don't pay for Day 1 and we extend support to 60 days at no charge.
Why us
Generative Specification is a DOI-backed, independently verifiable programming discipline. The workshop teaches a named, formal methodology — not tips from a YouTube channel. genspec.dev →
Senior architects with active production work in data architecture and software engineering. We write code every day. We know what breaks in real codebases at scale.
ForgeCraft and CodeSeeker are published, open-source tools with real distribution. Your team inherits tools that work in production — not workshop utilities that expire when the contract ends.
English and Spanish. Instructors based in Minnesota and Latin America. International reach, production-grade delivery standards.
30-minute discovery call. We assess your codebase, explain exactly what the two days look like for your team, and tell you honestly if it's a fit. No pitch deck. No pressure.
Book a discovery callOr email directly: juan@pragmaworks.dev
forgeworkshop.dev · genspec.dev · forgecraft.tools