AI Build Sprint · 10 business days
We build working AI with your team in 10 days.
Bring us one product or workflow, and our senior builders ship it with your people while your team learns to build the next one.
Book a 30-min call30 minutes. No prep. You talk to a builder, not a salesperson.
Our clients
Where our team has delivered
Why AI stalls
Most AI work stops short.
- The pilot works. Then it never ships.
- The consultants leave slides. Nothing gets built.
- The tool shows up. Nobody on your team can run it.
We build the real thing, with the people who will run it.
What you get
On day 10, you have three things.
On the Governance and People tracks, the thing you get is a working policy or programme, not software.
A working product, in use.
Real users put it to work before we leave.
A team that can build the next one.
Your team builds with us every day. They keep going after we go.
The code and docs. Yours.
You keep the code, the docs and the keys. Nothing to renew with us.
The 10 days
Ten days from kickoff to hand over.
Day 1
Scope
We lock what we build and what “working” means.
Days 2 to 8
Build together
We build with your team every day. You see it grow.
Day 9
Real users try it
The people who will use it put it to work. We fix what they find.
Day 10
Hand over
You get the product, the code and the docs. Your team runs it from here.
Four tracks
Pick the kind of work. Meet who leads it.
Product
Take a prototype to production, or build a new product.
Led by John Ellison, with Chetan Nandakumar
Workflow automation
Agents take the repeat work. Your people stay in charge.
Led by Abe Isleem
Governance & policy
Clear AI rules your teams can follow, and people ready to approve new AI use.
Led by Stephan Ledain
People & culture
AI fits the work your teams already do, and they have the skills to use it.
Led by Fatma Ghedira
Work we've shipped
Real builds. Real numbers.
Insurtech
of calls now reviewed
A UK insurtech: in two weeks we built two working prototypes, AI call reviews and a knowledge assistant.
Read the case studyReal estate
estimated saving per trained person, per week
A European real-estate firm: we built a knowledge assistant, then trained 7 of their people to build.
Read the case studyIs this for you?
For you if:
- You already use AI and want a real product from it.
- You have one product or workflow in mind.
- A leader can make calls fast.
- One person on your team can build with us each day.
Not for you if:
- You want a slide deck or a strategy memo.
- You want AI training with nothing built.
- Five teams need to change at once.
- We can't get access to your tools and data by day 1.
The team
The people who build with you.
16 years building software. Built and exited two companies.
AI PhD from UC Berkeley. 10+ years launching AI products.
Senior AI engineer, 8+ years, formerly at Microsoft.
EU AI Act specialist. Founder of AdaptAI.
Organizational psychologist. Has worked with Google, Snapchat and the World Economic Forum.
Pricing
One fixed price, agreed before you start.
Product sprint from $20,000.
Fixed price agreed before you start.
Book a 30-min callNot ready for a sprint? Workflow Triage: 2 to 3 days, from $3,000, credited to a sprint you book within 30 days.
After the sprint: retainers from $3,000 a month.
All prices in US dollars.
Not ready to talk? Take the free AI readiness scorecard.
Questions we get asked
How much does it cost?
Sprints start at $15,000. A Product sprint starts at $20,000. We agree one fixed price before you start, so there are no surprises.
Who owns what you build?
You do. You keep the code, the docs, the setup and the logins. We list anything run by a third party, and what it costs, before you sign. You never have to keep paying us.
How do we agree what “working” means?
Before you sign, we write down what it must do and who checks it. On day 1, we confirm that list with your team. On day 10, we check the work against it together.
How much of our team's time does it take?
One person from your team builds with us each day. Your sponsor joins on day 1, for key calls, and on day 10. You see the hours for each role before you sign.
Is our data safe?
We only use data you approve. Before we start, we list every AI model and service that will touch it, and where it runs. If we can't meet your security rules, we won't take the sprint.
What if it isn't done by day 10?
We agree the scope before you sign and keep it small enough to finish. If something we agreed isn't working, we fix it inside the sprint. You never have to buy more work to finish what we agreed.
What happens after day 10?
Your team runs it. Thirty days later, we meet for an hour to see how it's going. After that, you can stop, or keep us on a monthly retainer. It's your call.
Has anyone else run this format?
Yes, Uber. Its Agentic Pods pair an AI engineer with a finance, legal, HR or marketing expert for two weeks to build together. Uber isn't our client. Our sprints follow the same shape.
Read the Uber CTO's post
“Tell me what should work differently in 10 days. I’ll tell you if we can build it.”
Image credit: Hourglass with sand by John Morgan, CC BY 2.0, graded from the original.