AI Discovery Sprint
The problem
Many teams want to use AI but do not know where it actually creates value, or they chase the flashiest use case without knowing whether it is viable or whether it pays off. The result: months and budget burned on the wrong bet, or a pilot that should never have started.
What makes a use case worth it
Not everything you can do with AI is worth doing. A good use case:
- Creates measurable value: it moves a number you care about, not a pretty demo.
- Is viable with your data: you have what the case needs, or a clear path to it.
- Tolerates its failure mode: you know what happens when it is wrong and you can bound it.
- Fits your budget: cost and latency work at your real volume.
- Has someone to use it: someone adopts and owns it, it does not end up orphaned.
How I work
- Evidence before enthusiasm: I prioritize by data and feasibility, not by hype.
- Vendor- and framework-agnostic: I recommend for your case, not for my favorite tool.
- Honest about feasibility: if AI is not the answer, I tell you.
- No lock-in: the blueprint is yours and anyone can execute it, with me or without me.
The process, in phases
A time-boxed way of working. It adapts to your context.
- Immersion. I understand your business, your data and where it hurts, with your people.
- Opportunity mapping. I identify where AI can move the needle, concretely.
- Prioritization. I score each opportunity by value and feasibility, and we choose the bet.
- Feasibility test. I validate the chosen bet with spikes or a light POC before committing to a build.
- Blueprint. I hand you the path to production: architecture, risks, cost and plan.
Deliverables
An actionable blueprint, not a PDF that gets filed: the prioritized opportunity map, proven feasibility for the chosen bet, a view of the proposed architecture (C4 model) and the key decisions on record (ADRs), with their cost and a plan to production. Ready for your team to execute.
A blueprint made system: an AI agent for regulated industries
Who it is for
For teams that want to use AI and need to decide where and how before investing in a build. For founders and leaders who have budget but not a roadmap.
Has your pilot already shown value and all that is left is getting it to production reliably? That is the next stage: the POC-to-Production Sprint. Not sure which one you need? Email me and I will point you the right way.
Frequently asked questions
What is an AI Discovery Sprint?
It is a time-boxed engagement to find where AI creates real value in your business and validate it is viable before building; it ends in a production blueprint, not a demo.
What does it mean for a use case to be viable?
That it can reach production with your data, within your cost and latency budget, and with a failure mode you can bound. I prioritize by evidence and feasibility, not by hype.
How is it different from the POC-to-Production Sprint?
The Discovery decides what to build and why; the POC-to-Production Sprint takes a pilot that already proved value to production reliably. The Discovery comes first.