Why A Sprint Beats A 6-Month Consulting Project
Six-month AI strategy projects produce binders. Sprints produce decisions. A 30-day AI discovery sprint forces the team to pick three real use cases, quantify their opportunity, run a small test of the riskiest one, and commit to a 90-day build. By day 30 you have either: shipped your first AI win, or knowingly chosen not to. Both are progress.
The sprint structure also forces decisions you'd otherwise defer: who owns AI, what the budget actually is, what the team is willing to change. Those decisions don't happen in consulting decks; they happen under deadline.
Week 1: Map
Shadow operators across functions. Document where time goes. The deliverable at end-of-week is a one-page heatmap of the business: which workflows consume the most time, where humans are doing AI-suitable work, and which workflows are politically untouchable.
Pick exactly five candidate use cases. Not three (too narrow), not ten (too unfocused). Five forces real prioritization.
Week 2: Quantify
For each of the five candidates, attach numbers: hours per week, dollar cost per year, ROI if AI handles 60%, ROI at 80%, what could go wrong, who champions internally, who resists. The numbers are estimates — that's fine. The discipline is the point.
- • Current hours/week consumed.
- • Fully-loaded cost.
- • AI deployment cost.
- • Implementation timeline.
- • Adoption risk.
Week 3: Test
Pick the riskiest of the top three. Build the smallest version that proves (or disproves) the AI approach can handle it. This is not the final product — it's a probe. Time-box it strictly.
The test usually answers one question: does the AI quality meet the bar your users will tolerate? Everything else (UX, scale, integration) is a separate concern.
Week 4: Decide
One half-day workshop, all senior stakeholders. Review the numbers. Review the test results. Pick the use case to build. Define the next 90 days. Name the owner. Allocate the budget. Set the success metric.
End-of-sprint deliverables get signed by the CEO. Without the signature, the sprint was theatre.
The Four Deliverables
- The Map. One page showing where AI fits and where it doesn't.
- The Quantification. Five use cases with numbers.
- The Test Report. What the probe revealed.
- The 90-Day Plan. Scope, timeline, owner, budget, success metric.
What Not To Do
- Don't hire a Head of AI before the sprint. Decide the role based on what you learn.
- Don't extend the sprint. The deadline is the discipline.
- Don't outsource the entire sprint. The founder must be in the room.
- Don't pick five use cases that all flow through the same team.
A 30-day AI discovery sprint produces a real plan. A 6-month AI consulting engagement produces a beautiful PDF. The difference shows up in your P&L a year later.
See our SMB roadmap.
FAQ
Should we hire a consultant? A specialist for week 3's test build, fine. The strategy work belongs to operators.
What if our five use cases are all bad? Real possibility. Better to know in 30 days than 6 months.
How big a team for the sprint? 3–5 people, including the CEO part-time.