AI Product Sprint
From idea to working alpha/MVP in 8 weeks
A focused, end-to-end sprint for teams who want to ship something real – not spend a quarter debating what “AI strategy” means.
Who this is for
This sprint is a good fit if you’re:
a startup building an AI-powered product or service
an organisation looking to automate or augment cumbersome workflows with AI
an innovation team that needs stakeholder alignment and_ something tangible to prove value
You’ll get the most value when the problem is high-impact and slightly messy – the kind that needs both design thinking and technical judgement.
Who it’s not for
This isn’t ideal if you want:
a “prompt pack” or a tool recommendation report
a long discovery phase ending in a polished 100-page presentation but no prototype
a magic ML model that removes the need for product decisions, data decisions, and governance decisions (tragically, physics still applies)
You’ll work with me directly, typically alongside a small team from my Creative Crow collective. I lead the sprint and stay hands-on throughout, while bringing in trusted collaborators (product design and engineering/creative technology) to move quickly without cutting corners.
You’ll leave with: a tested, working prototype (and often a working alpha), a clear product definition with best-practice UX and UI patterns, an evaluation plan, and a delivery-ready roadmap.
What you get
The sprint is designed to produce a product-focused tangible prototype or alpha – something your team can build from, test, and take into production.
Typical outputs include:
Opportunity brief – what problem we’re solving, for whom, and why now
Product definition + scope boundaries – what the AI does, what it doesn’t do, and where humans stay in the loop
Working prototype(s) – a functional prototype for the critical path (and often smaller POCs for individual components)
User testing & validation plan – testing with real users, a validation framework for ongoing iteration, and metrics for usability and product–market fit
Risks & assumptions log – what could break, what we’re betting on, and what to validate next
Delivery-ready roadmap – what to build next, in what order, and why
Depending on the project, we may also deliver:
a simple, AI-tailored design system
a lightweight data readiness assessment
tooling / architecture recommendation for the MVP phase (vendor-neutral)
a short exec-ready narrative deck (short, crisp, decision-focused)
How it works
The “8 weeks” aren’t seven identical weeks – it’s a sequence designed to reduce risk early and deliver value iteratively.
You’ll work with a small team on our side (typically me as lead, plus a product designer and an engineer/creative technologist). We run weekly check-ins, plus a handful of collaborative sessions for alignment, critique, and decision-making.
Week 1: Discovery
We build a solid understanding of the problem, the business need, and the users – without focusing on AI or tech just yet.
Week 2: Setup & initial alignment
We sketch early concepts, align on the critical path, and set up the prototype foundation (tooling, hosting, repo, and deployment) so we can iterate quickly. You will have a link to a live prototype from this week on.
Weeks 3–7: Designing, prototyping, and testing
We design, prototype, and test the key building blocks. This is where we turn “AI idea” into a product flow and pressure-test it with real scenarios and users.
Week 8: Documentation & next steps
We consolidate what we’ve built and learned into a handover that your team can run with: product definition, prototype/alpha code repository, Figma designs + slides, evaluation plan, and a clear roadmap.
What makes this sprint different?
It’s product-led, not tech-led
We start with the user outcome and the workflow – then choose the AI approach that fits. Not the other way round. However, there will be a whole lot of geekiness involved.
You get something real
Not a slide deck presentation. You’ll have a prototype your team can test with users and stakeholders – often a working alpha.
It’s fast because it’s disciplined
Speed comes from a small, highly focused team, tight feedback loops, and making decisions early with only the right people in the room.
What we need from you
To run the sprint well, we’ll ask for:
A single accountable point person (product owner or exec sponsor)
Access to 2–4 domain experts (the people who know what “good” looks like and can help us validate ideas quickly)
Access to end-users for testing (typically 3–8 users, depending on scope and availability)
A technical counterpart (or access to engineering for feasibility checks)
Data access (a subset of real data or representative sample data)
A weekly cadence for reviews and decision-making
Time commitment (typical):
Sponsor: 2–3 hours/week
Domain experts: 1–2 hours/week
Technical counterpart: 1–2 hours/week
If your calendar is already on fire, we can still do this – but we’ll need to be deliberate about who shows up and when.
What this tends to unlock
Examples of sprint-style outcomes:
collapsing a manual process from days to minutes
transforming a complex process into a streamlined workflow
making an AI system’s reasoning legible so teams can trust and debug it
turning “we should use AI” into a scoped product with an evaluation plan and a buildable roadmap
Want specifics? See the case studies.
Frequently asked questions
-
A one or two-weeks sprints are great for simple idea generations and quick testing. Our goal is different: to deliver a more robust prototype (and often a working alpha) that covers multiple parts of a real product or service.
We spend Week 1 properly understanding the problem and the users, then we take the time to design, build, and test deeply enough that the outputs are genuinely useful – not just plausible.
-
To move quickly we typically use our own proven setup (often Nuxt/Vue plus Supabase/InstantDB/Upstash depending on constraints). We usually work inside your environment from day one – for example using your GitHub for version control and your Vercel/Netlify/Cloudflare for deployment – so you have access to everything as we build it.
-
That’s valuable learning. The sprint is designed to fail fast and cheaply – better to discover fundamental issues in Week 3 than after 6 months of development.
If the core AI approach hits a wall, we’ll pivot to alternative approaches or reframe the problem. You’ll still come away with a clear product definition, validated user insights, and a realistic roadmap based on what’s feasible with current technology.
-
Typically, we hand over the prototype/alpha to your team with clear documentation, decision context, and next steps so you can move forward independently.
Many teams benefit from periodic check-ins as they go from prototype to production – we can discuss light-touch advisory support during the sprint scoping call, but it's not required to get value from the core 8-week sprint.
In some cases, we’ve also stayed on to help build and ship the production version. If that’s something you’re considering, we can talk through options and fit on the scoping call.
-
No – but we do need representative data or realistic test scenarios. Week 1 typically clarifies what we can do now versus what needs investment.
-
Tooling is selected based on your constraints: security, cost, latency, data residency, observability, and integration needs. Our tech-stack can be deployed anywhere and the sprint stays vendor-neutral unless you have a strong preference.
-
We treat these as product constraints from day one. If you have specific policies, we design within them.
-
That’s common. We can discuss this on the scoping call. The sprint helps you pick a high-value slice and validate it quickly – without boiling the ocean.
-
No. The sprint is a catalyst – it gives your team a clear product direction, working artefacts, and a plan they can execute.
Book a Sprint Scoping Call
On the call we’ll:
map the opportunity and constraints
pick a viable, high-value slice for 8 weeks
agree what “success” means
decide whether the sprint is the right fit