FOR COMPANIES

Give them somewhere to build inside.

Your best people are already building things in the evenings. Some of them will leave to do it properly. Twelve weeks, twenty employees, and every one of them ships something real.

Talk to us
TWELVE WEEKS · TWENTY EMPLOYEES · EVERY ONE SHIPS
01 WHAT IT COSTS YOU NOW

Three quiet losses.

01
The idea leaves with the person
The employee who has been building at night does not raise it at the town hall. They hand in notice eighteen months later, and the thing they built becomes somebody else's company — usually one that sells into your market, using what they learned inside your walls.
02
The AI capability never lands
Training gets watched, not used. People finish a course about a tool and return to work that never asks them to open it. Capability only sticks in people who have shipped something with it, and almost no programme requires that.
03
Every prototype goes to procurement
An idea needs a working version before anyone can judge it. Getting one means a spec, a vendor, a quote and a quarter — so most ideas are judged as documents, and the cheap ones die beside the expensive ones.
None of these is an idea problem. You have plenty of those. THEY ARE ALL THE SAME BOTTLENECK: NOBODY INSIDE COULD BUILD.
02 — WHAT CHANGED

One person can now build real software.

Directing AI, without writing code and without hiring anyone. This is not a projection. The platform this programme runs on — the applications, the curriculum delivery, the assessments, the participant tracking — was built by one person, alone, directing AI, for about 3,250€ in tooling. You are reading this on it.

3,250€
in tooling
1
person
The platform this programme runs on.
03 — WHAT THAT CHANGES INSIDE A COMPANY
01 The people with the problem can build the answer

The person who knows exactly where the process breaks is usually not the person who can build software. That gap is what every internal innovation programme has been routing around. It has closed.

02 Capability arrives as a side effect, not a course

Nobody who has shipped a working product with AI needs convincing about it afterwards. Twenty employees who have done it once are a different organisation from twenty who attended a workshop.

03 An idea can be tested before it needs a budget

A working version in twelve weeks, built by the person who proposed it, costs less than the meeting series required to approve a vendor.

WHAT THE COMPANY GETS

Outcomes, not hopes.

Working software, counted

A live URL per participant is a binary fact, not a judgement.

AI capability that actually stuck

Because they built with it rather than watched a course about it.

A retention answer

For the specific people most likely to leave, and the hardest ones to replace.

Ideas tested cheaply

Twenty of them, on evidence, in a quarter — with the ones that do not work identified early and honestly.

Internal tools you keep

Where you choose that arrangement, the products belong to the company.

Reporting, per participant and per cohort

From the platform. Nothing assembled by hand for the steering committee.

05 — WHO SHOULD BE IN THE ROOM

Not the volunteers. The ones already building.

The people who put their hand up for an innovation programme are rarely the people quietly shipping side projects at eleven at night. The second group is who this is for: deep domain knowledge, a problem they have watched go unfixed for years, and enough restlessness to have already started doing something about it on their own time.

07 — TWELVE WEEKS · TWENTY EMPLOYEES · ONE PRODUCT EACH
1–2 Problem and idea

A written problem statement and a scoped idea.

3–4 Buyers and validation

Twenty named buyers — internal or external — and five documented conversations.

5–7 Build

A working end-to-end product.

8–9 Money and readiness

Payment or access flows live, pre-launch audit passed.

10–11 Launch and users

A live URL and real users giving feedback.

12 What next

A business case and a 90-day plan.

THE COMMITMENT

Ten to twelve hours a week.

One 90-minute live session, recorded, and small peer groups between sessions. It is built to run alongside a full-time role rather than instead of one. There is no technical prerequisite — participants do not need to code, and they are not taught to. They are taught to direct.

08 — WHO OWNS WHAT GETS BUILT

Decide this before week one, not after.

The company owns it
SIMPLEST

Standard work-product terms. Cleanest to administer, and the right answer when the brief is internal tooling. Participants should be told plainly, at the start, that this is the arrangement.

They own it, you get first refusal
THE ONE THAT KEEPS PEOPLE MOTIVATED

The employee owns what they build; the company takes a licence to use it and the right of first refusal if it becomes a business. You get the tool and the capability, they get a reason to do the hard weeks, and neither side is pretending.

It spins out
RARE, BUT WORTH HAVING A ROUTE FOR

For the small number of products that clearly outgrow their department. Terms agreed in advance so the conversation is a process rather than a negotiation held under pressure.

WHICHEVER YOU CHOOSE, IT IS WRITTEN DOWN AND TOLD TO PARTICIPANTS BEFORE THEY START. AMBIGUITY HERE IS WHAT TURNS THESE PROGRAMMES INTO RESENTMENT.

THE SPLIT

What we provide, and what you provide.

We provide — the curriculum

The full twelve-week curriculum and every weekly deliverable.

We provide — the platform

Applications, selection, curriculum delivery, participant tracking and reporting.

We provide — the live sessions

Delivered by someone who has built and shipped the thing being taught.

We provide — facilitator certification

Training and certification for your own people, so you can run later cohorts internally.

You provide — the participants

Twenty employees, selected with us rather than volunteered.

You provide — the IP decision

One of the three arrangements above, agreed before week one.

You provide — protected time

Ten to twelve hours a week that their line manager has actually agreed to.

You provide — the room

For demo day, if it is held in person.

10 — WHERE THIS IS TODAY

No company has run this yet.

The consumer cohort is the first one and it has not run. There are no corporate references and no case studies, and I am not going to invent any. What exists is the method, the platform it runs on, twenty years of building products inside organisations, and one worked example: mine. The first company to run this is buying a pilot, and should be priced like one.

11 — WHO IS BEHIND IT

Luis Gonçalves

Twenty years in digital product development. Led the agile transformation at Nokia. Teaches at Porto Business School. Before this, a venture that lost about 250,000€ — which is where the method comes from. Then the platform you are reading this on, built alone, directing AI.

3,250€
in tooling
20 yrs
in product
?12 — THE OBJECTIONS
How is this not another innovation programme that produces nothing?
Because the deliverable is not a recommendation. Every week ends in something that exists, and week twelve ends in a deployed product per participant. If that does not happen, the programme visibly failed — which is the point. Most innovation programmes cannot fail visibly, and that is why they persist.
Do our people need to code?
No, and they are not taught to. They are taught to direct.
Who owns what they build?
You choose one of three arrangements before week one, and participants are told which one applies. See the section above.
Is this not just teaching people to leave?
The people this is aimed at are already building at night, and the ones who leave do it whether or not you run this. What changes is whether the thing they build happens inside a structure you can see, with your problems attached to it, and whether they end the year feeling backed or overlooked.
What about our security and data policies?
Participants build on their own accounts unless you specify otherwise, and the brief can be scoped to exclude production data entirely. This should be agreed with your security function before selection, not after.
Can our own staff run later cohorts?
Yes. They are trained and certified first, and they do not need a technical background.
How do you select the twenty?
With a selection instrument, and with you. We select for domain expertise, a problem seen up close, and time a line manager has genuinely protected. Not technical ability, and not enthusiasm in an application form.
What if someone's product does not work?
Then you know inside twelve weeks, on evidence, for the cost of some evenings rather than a vendor engagement. That is a result, not a write-off.

They are going to build it either way.

The only question is whether it happens inside a structure you can see, with your problems attached to it — or eighteen months from now, in a company that competes with you.

Talk to us How it works, in full
PRICED PER COHORT, NOT PER SEAT · IP TERMS AGREED BEFORE WEEK ONE