3–5 weeks.We transform 20–50-person engineering teams into AI-native delivery organisations. Measurable productivity uplift. No need to hire for the next 12–18 months. The spec is the contract — yours, not a template.
The market shifted. Most teams haven't. They're either still writing tickets and treating Copilot as autocomplete, or they handed engineers Cursor and Claude Code with no governance and called it transformation. Both fail.
You bought the licenses. You wrote a memo about AI being a strategic priority. The senior engineers use Copilot for autocomplete and the juniors use ChatGPT in a separate tab. The SDLC didn't change. The artifacts didn't change. The throughput didn't change. You're paying for AI and getting tickets.
You handed every engineer Cursor and Claude Code. Velocity spiked. So did the production incidents, the unreviewed PRs, and the codebases nobody understands. Two senior engineers are quietly job-hunting because the work isn't engineering anymore. You're shipping faster, breaking more, and your CTO can't explain what changed.
▸ The fix isn't more tools or more memos. It's a workflow built for AI-native delivery — and the artifacts to govern it. That's what we install.
A senior engineer in our markets costs $80–120k fully loaded, takes 3–6 months to hire, takes another 3–6 months to ramp, and there's a 30–40% chance they leave inside 18 months. Every CTO at this size is trapped in a hiring loop that eats their calendar and their runway.
Three to five weeks of intensive transformation, and your existing engineers produce what twice as many would have. You don't need to hire for the next year. When you do, you hire for capability gaps — not capacity.
Audit, design, workshop, supervised pilot, measured showcase. The training is one component — step 03. The other four are why this works when "buy 30 seats of a course" doesn't.
We map how your team actually delivers — gates, artifacts, ceremonies, decision points, where time goes. We talk to engineers, not just the CTO. By the end of week one, we know more about how your team delivers than you do.
We design the new workflow specifically for your org. Not a template. Your team, your stack, your constraints, your compliance environment. Nine years of delivery scar tissue and three published essays of methodology become a custom playbook with your name on the cover.
Every engineer in the org goes through the same training that runs in our senior D2C programme — but tailored to the workflow we designed in step 02. PMs and engineering managers get a parallel track on spec-first leadership.
We pick one of your teams and run a real sprint using the new workflow, with our supervision. This is where the methodology meets your actual codebase. We watch, intervene where needed, document what happens. The sprint produces shipped work, not theatre.
We measure the pilot sprint against a baseline sprint from before the engagement and present the delta to leadership. Throughput, review cycles, time-to-merge, scope held. This is the artifact your CTO takes to the board.
The engagement is calibrated for a specific shape of company. If you're outside that shape, one of the other doors at specshop.dev is probably better — or we'll tell you to call someone else.
Every clever CTO reads the method, gets to step 03, and asks this. We'd rather answer it on the page than dance around it on the call.
The training is one component of a larger engagement. Without the audit (step 01), the custom workflow design (step 02), the supervised pilot (step 04), and the measured outcome (step 05), you have engineers with new ideas and an old process. We've watched it happen at three different companies that bought our content as a course. The methodology bounces off the existing workflow and the team reverts inside a quarter.
The reason transform works is that we don't drop a methodology onto your team and wave goodbye. We rewire the workflow it lands on. Then we run a real sprint inside that new workflow with our supervision, and we measure the delta against a real baseline. Without those four steps, the workshop is hope. Hope is not a transformation strategy.
We're not a consultancy theorising about AI transformation. We're a delivery practice that lived it, built the methodology, published the thinking, and now installs it in other organisations.
Specshop is the spec-first successor to Beta Launch — the agency we ran for nine years across five countries. The methodology isn't theory. It's scar tissue.
Before we sold transformation, we did it on ourselves — capped the team under 10, rewired the workflow, shipped against the new method. The numbers came from inside the building.
Three foundational essays in the journal lay out the method in detail. Read them before you call. If the argument doesn't land, the engagement won't either.
Below is a representative engagement profile, not a specific client. We'll publish real measured outcomes as Q3 engagements wrap. The shape is real; the numbers are calibrated against our internal delivery transformation and the Beta Launch baseline.
Most consultancies hide prices. We won't. Our buyers are competent and time-poor — and the math works against the alternative without a sales process to defend it.
depending on size, complexity, and number of teams in the pilot scope
Book a discovery callPre-loaded here so you can come into the call ready to talk about your team, not the engagement structure.
The training is step 03 of a five-step engagement. Without the audit, the custom workflow design, the supervised pilot, and the measured outcome, you have engineers with new ideas and an old process. We've seen it bounce off real teams more than once. If you only want training, /train is the right door — and we'll point you there ourselves.
Two differences. First, we ship — we don't write decks. The engagement deliverable in step 04 is a real sprint of shipped work, not a slide. Second, we're a small senior team that did this transformation on our own delivery first. The team installing it has built software using the method for years, not weeks.
Engineers give us roughly four hours of interview time across week one, attend the workshop in week three (typically two days), and one team runs the pilot sprint in week four as their normal sprint cadence. Leadership commits to the showcase at the end. That's it. No homework, no parallel tracks, no all-hands re-orgs.
It will — but the more useful answer is what we do if it doesn't. The 30-day check-in is the second look. If the measured delta is below what we projected at the start of the engagement, we extend the supervised pilot at no additional cost until it lands. We've never had to do this. We've also never had a client we couldn't measure.
Probably not as a transform engagement — the economics don't work for either side at that size. The senior training programme on /train is built for engineers at smaller orgs who want to take the methodology back to the team. We can also point you at a lighter spec-only build engagement if that's a closer fit.
The team is split across Colombo, Muscat, and Dubai. We run the audit and workshop in your timezone — on-site where it makes sense, remote where it doesn't. The pilot sprint and the showcase are always run in your team's working hours.
We're capped at two transformation engagements per quarter. Q3 has two slots open as of this version of the page. The discovery call typically happens within a week of first contact; the engagement starts 2–4 weeks after the call.
A 30-minute discovery call. We'll tell you whether transform is the right shape, or whether one of the other three doors makes more sense. Either answer is honest.