Tech.Rocks 2026: Peinture kicks off the debate on the hidden cost of tech projects
Your company optimizes every tax it can, except one. On November 30 and December 1, come see a live demo of how to cut it.
Every time an idea moves from one team to the next, time gets lost. It shows up in no budget, yet every product team pays for it. We call it the translation tax, and here’s what it looks like in 60 seconds.
One simple question, a ticket going in circles
It starts on the business side: “Can we just see this month’s revenue on the dashboard?”
Then the ticket goes on a trip:
To Do → Refinement → Design → In Progress → Needs Clarification → Code Review → QA → Blocked
On the way, it picks up an “alignment sync” with seven guests. On the mockup, the main number becomes total revenue, and MRR (monthly recurring revenue) moves to the corner. The pull request (the proposed code change) comes back with changes requested. And an email thread keeps growing: “Re: Re: FW: Revenue number for the dashboard”.
Someone finally asks the real question: “Spec says MRR, mockup says total. Which one?” Meanwhile, Marketing just wants to know: “Any update on the revenue thing?”
Nobody did a bad job. Every handoff is a translation, and every translation has a cost.
90 hours for a single number
If someone sent the bill for that ticket, it would look like this:
| Translation tax notice | Time |
|---|---|
| Clarification meetings | 14 h |
| Rewritten | 6 h |
| Back-and-forth | 9 h |
| Wrong assumptions | 5 h |
| Rework | 18 h |
| Scope changes | 11 h |
| Handoff fixes | 7 h |
| QA ping-pong | 8 h |
| Delayed release | 12 h |
| Total due | Way too much (90 h) |
The figures are illustrative. Yours will differ, but the line items probably won’t.
Faster code won’t fix it
Look at the bill: writing the code isn’t on it. AI has already made implementation cheap (we covered that in Implementation Isn’t the Expensive Part Anymore). The hours get lost at the point where the team has to agree on what to build.
Generating code faster only gets the wrong version to QA (testing) sooner. This is the tax your company forgets to optimize, and cutting it means everyone starts from the same version.
Everyone in the same prototype
That’s exactly what Peinture is for. Business, product, design and engineering work in the same prototype, at the same time, on top of your existing code.
In the video, the business side types the request in plain language: “Add this month’s revenue to the dashboard, using our design system.” Peinture reads the dashboard and the design system (the company’s component library), then states its assumption: revenue means total invoiced this month, in euros, not MRR. The question that took a whole email thread gets settled at once, in front of everyone.
The card is built with a component that already exists. Design adjusts it on the spot. Engineering opens the Documentation tab: data source, display rules, states, interactions. It’s all written from the prototype the team just approved.
Peinture is built to be:
- Multiplayer: the whole team in one prototype. No more juggling four tools and an email thread.
- Sovereign: each prototype runs in an isolated space. You can also host everything on your own infrastructure.
- Flexible: it works with several AI tools and models, such as Claude Code, OpenCode, Microsoft Foundry, Mistral Vibe, Prisme.ai or your own model, in the cloud or on-premise.
Why we chose Tech.Rocks
Tech.Rocks is where CTOs and tech leaders meet, and they’re the ones who see the bill. For this 10th edition, the program covers AI in production and sovereignty, two topics at the heart of Peinture.
Many teams are now building an AI factory: AI agents writing code at scale. But an agent only builds well what it was explained well. Peinture sits just before that: the team agrees first, then the agents start from specs everyone has approved, we like to call it an AI Innovation Lab.
So, are you ready to uncover the translation tax with us?
Come find us at the booth
- Who: Ivan Dalmet and Rudy Baer
- Where: Théâtre de Paris, Start-Up Pack area
- When: November 30 and December 1, 2026
Stop by between two talks for a live demo of Peinture.
Can’t make it to Paris? We’ll run the same demo remotely.
Request a demo