A real prototype, built with AI, tested by your customers in 3 days
Not a mockup: a working application, with your data and, if it’s relevant, real AI features. Your team chooses the solution, AI builds it in one day, and 5 real users test it the next day.
Prototyping with AI
What we build in one day
With AI, a prototype is no longer a picture pretending to be a product. Your testers use it like a real product, and that is what makes their reactions reliable.
A web or mobile app
Screens that respond, forms that save data, a complete journey to run from start to finish.
An AI assistant or agent
It gives real answers, based on your documents. Your testers ask their own questions, with no script.
A dashboard on your data
Your metrics, on an anonymized extract of your real data, to see what actually helps you make decisions.
A purchase or sign-up journey
Order, quote, onboarding: we test the entire journey, including the cases where it gets stuck.
An internal tool
The form, workflow or tool your teams keep asking for, tested by the people who will use it.
An offer and its message
An offer page and its message variants, to measure what creates demand before you build.
The problem
Six months of development to find out nobody wanted it
Meetings go in circles
Everyone defends their own idea, the loudest voice wins, and the topic comes back at the next committee.
You build before you verify
The need is assumed, rarely tested. Customer feedback arrives after go-live, when changes are expensive.
AI promises a lot, but what should you build?
Everyone wants to « do AI ». Few teams know which use case is truly worth the effort.
The classic Design Sprint is intimidating
Five days with the decision-maker and your best people locked in: hard to fit into a calendar.
The method
What is a Design Sprint?
A Design Sprint is an innovation process that lets a small team (a decision-maker, designers, developers, domain experts) go from a problem to a prototype tested by real users in just a few days. A Design Sprint lets you test new ideas and reduce the risk of an innovative product, service or feature before you commit months of development.
The origins of the Design Sprint
The method was invented by Jake Knapp at Google, then refined at Google Ventures (GV) with John Zeratsky and Braden Kowitz, before being popularized by their book Sprint (2016). Since then, hundreds of product teams, from startups to large corporations, have used it in workshops to make decisions fast.
The Design Sprint methodology in 5 steps
The classic Design Sprint takes 5 days, one step per day:
Map
Long-term goal, journey map, expert interviews, choice of target.
Sketch
Everyone draws their own solution, alone and in detail, building on the best existing ideas.
Decide
Vote, decision by the decision-maker, then a storyboard of the experience to test.
Prototype
A realistic prototype, just real enough to get genuine reactions.
Test
5 user tests, in one-on-one interviews, and the lessons that come out of them.
Design Sprint or design thinking?
Design thinking is an approach: understand user needs, define the problem, ideate, prototype, test. The Design Sprint is a time-boxed way of putting it into practice, with a strong constraint (a few days), a decision-maker in the room and a decision at the end. It also shares the spirit of Lean Startup: test before you build.
Why run a Design Sprint?
- Make a decision in days rather than months of committee meetings
- Test new ideas without tying up a team for months
- Reduce risk before investing in development
- Align business, product, design and tech on a single goal
- Learn from your users, not from your assumptions
- Speed up the launch of what works
Our AI-accelerated version keeps the essence of the method (the decision-maker, the individual sketches, the vote, the 5 tests) and compresses the rest, mainly preparation and prototyping. The result: 3 days instead of 5, and a working prototype instead of a mockup.
The schedule
What the team does, what the AI does
Prepare
👥 The team
- Frame the challenge with the decision-maker
- Choose the team and the target
- Recruit 5 users for day 3
🤖 The AI
- Summary of what you already know: customer verbatims, support tickets, reviews, usage data
- First overview of the competition
→ A clear challenge and a starting file that the whole team has read
Understand, sketch, decide
👥 The team
- Long-term goal and sprint questions
- Journey map and choice of target
- Individual sketches, in silence, before any AI
- Vote, then the decision-maker decides, and storyboard
🤖 The AI
- Transcription of expert interviews and grouping of the « How might we… » notes
- Lightning demos: what is done elsewhere, in ten minutes
- Sparring partner to push the sketches further
→ A chosen solution and a storyboard ready to prototype
Prototype with AI
👥 The team
- The team describes, tests and corrects as it goes
- Business experts check every screen and every answer
- In the evening: a dry-run test and the interview script
🤖 The AI
- An application that really works, not a mockup
- Your data (anonymized) and real AI functions when AI is the subject
- Two variants when the team can’t decide
→ A working prototype, ready to be put in your customers’ hands
Test and decide
👥 The team
- 5 interviews with real users, observed by the whole team
- Reading the results together
- Decision: go, pivot or stop
🤖 The AI
- Live transcription and note-taking
- Summary of recurring signals, backed by quotes
- Draft plan for the next steps
→ A decision based on what your users did, not on what you thought
What changes
The classic Design Sprint, only faster
| Classic Design Sprint | Design Sprint with AI | |
|---|---|---|
| Duration | 5 days, the whole team tied up | 3 days, plus a half-day of preparation |
| What you already know | Often ignored for lack of time | Synthesized by AI before day 1 |
| Prototype | A clickable mockup, with screens that only pretend to work | A working application, with your data and real AI features |
| What testers experience | They imagine the product | They actually use it |
| Test analysis | Post-its and the team’s memory | Live transcripts and synthesis, backed by quotes |
| What does not change | Humans decide | Humans decide, AI does not vote |
What is it for?
Four situations where a sprint changes everything
A new product or a new offer
Check that there is a real need before committing to six months of development.
An AI use case
Choose the AI agent or assistant that is truly worth building, and test it.
A journey that gets stuck
Sign-up, ordering, onboarding: understand where your customers drop off and test an alternative.
An internal process
A tool or an organization your teams struggle with: rethink it with the people who use it.
What you get
At the end of the 3 days
- A working prototype that you keep, tested by 5 users
- The recordings and summary of the 5 tests
- A clear, well-argued decision: go, pivot or stop
- The sprint report within 48 hours
- The hypotheses to track and the indicators to measure them
- The sprint prompt kit, reusable by your teams
The team
4 to 7 people, including the decision-maker
The decision-maker, plus people from product, design, tech, business or sales, and support. The decision-maker must be present at least on day 2, when decisions are made.
The facilitator
Robin Béraud-Sudreau started out as a UX designer and Scrum Master, followed by 15 years of agile and product coaching. He builds his own products and tools with AI, and trains teams and managers in agentic AI. His background →
They trusted us
Formats
Three ways to run your sprint
Facilitated sprint
Robin leads the sprint with your team, on your challenge.
On quote
- Remote preparation and scoping
- 3 days of workshop, on site or remote
- The working prototype, which stays yours
- Recruitment of the 5 testers (optional)
- Sprint report within 48 hours
Sprint + facilitators
The sprint, then one day so your teams can run sprints on their own.
On quote
- The full facilitated sprint
- 1 day of training for your future facilitators
- The complete kit: agenda, materials, prompts
- A follow-up check-in at 30 days
Train your facilitators
For product, design or transformation teams who want to adopt the method.
On quote
- The method, day by day
- Prototyping with AI: method, tools, limits
- A mini-sprint in real conditions
- The complete kit
Prices exclude VAT. Travel expenses and tester fees are extra, on quote. The full offer in 2 pages: the PDF brochure →
FAQ
Frequently asked questions
How long does a Design Sprint take?
The classic format, described by Jake Knapp in Sprint, lasts 5 days. Our Design Sprint with AI fits into 3 days, plus a half-day of remote preparation.
How do you prepare a Design Sprint?
With the decision-maker: we define the challenge, choose the team, book the 3 days and recruit 5 users for the tests. AI synthesizes what you already know so everyone starts from the same file.
What happens after the sprint?
You receive the report within 48 hours. If it’s a go, you know what to build first and what to measure. If it’s a pivot, a second sprint often lets you test the new direction.
Why 3 days instead of 5?
Because AI takes over what used to slow down the classic sprint: summaries, transcriptions, the prototype, test analysis. The moments when the team thinks and decides stay the same.
What exactly does the AI do?
It summarizes, transcribes, suggests variations and builds the prototype. It decides nothing: the sketching starts without it, and the decision-maker makes the call. A brainstorm that starts with AI produces average ideas, so we do the opposite.
Do you need to know how to code or master AI?
No. The facilitator handles the tools. Your team brings its knowledge of the business and of your customers.
Can the prototype be put into production?
No, and that’s not its purpose: it is built to learn fast. But it really works, you keep it, and your developers can pick up from what worked with your users.
Why a working prototype rather than a mockup?
Because a tester using a real product behaves like a real customer. Faced with a mockup, they imagine; faced with a prototype that works, they act. And for an AI topic, it’s the only way to test the quality of the answers.
What about the confidentiality of our data?
We use the tools approved by your IT department, or anonymized data. We discuss it during scoping, before any file is exchanged.
Who should take part?
4 to 7 people: the decision-maker, plus product, design, tech, business or sales, and support profiles. The decision-maker must be there on day 2 at the latest.
On site or remote?
Both work. On site, the energy is better; remotely, we use an online whiteboard and the format stays the same.
What if the test shows the idea doesn’t work?
That’s an excellent result: you learn it in 3 days instead of 6 months. Day 3 ends with a decision, including the decision to stop or pivot.
Have a challenge worth 3 days?
Tell us about it in 30 minutes. We’ll tell you frankly whether a sprint is the right answer, and if not, what would be.