8–12 bet cards
Each with a hypothesis, assumptions, gate and tier, plus the cheapest way to get proven wrong.
Fresh Kiwi · AI consultancy
Most bets fail. That's the design.
Opportunity AI doesn't sell a forecast; it lowers the cost of finding out. A typical round of cards produces 1–2 bets worth a probe, and the rest get killed on purpose, before any money moves. A probe is billed in full whether it succeeds or fails. The engagement itself runs through probe design for 1–2 bets; running the probes is not included. You get the prices in the qualification call.
Perspective
Why the audit's method doesn't work here
An opportunity hypothesis has no baseline, no volume, no observed task. Everything is assumed by construction. Run it through the Efficiency AI Audit's method and it does exactly what it's built to do: every opportunity gets capped at 'needs further investigation', every time. The method isn't broken; it's correctly refusing to score something it wasn't built to score. So an opportunity hypothesis gets no score. Instead, a bet card: the hypothesis, who pays and who uses it, the two or three load-bearing assumptions, the cheapest way to get proven wrong, and a kill criterion fixed before the probe starts, not written after the fact.
Overview
How it runs (about three weeks)
| Step | What happens | Duration |
|---|---|---|
| 1. Scoping call | Sponsor, appetite, and the portfolio budget: what is it worth to find out you're wrong here? | 30 min |
| 2. Thought-starter localisation | Desk work: the generic card deck rewritten for this sector and this business. | ~1 day |
| 3. Divergent session | 120 minutes with 8–15 people, deliberately wide: commercial, delivery, someone who talks to customers daily. | 2 hrs |
| 4. Bet drafting | Desk work: raw input into 8–12 bet cards, each with a gate, tier and the cheapest way to disprove it. | ~1 day |
| 5. Convergent session | 120 minutes with management, a week later. Most bets get killed; 1–2 probes get funded within budget. | 2 hrs |
| 6. Probe design | Probe sheets for the funded bets, with kill criteria signed off before anything starts. | ~0.5 day |
Capabilities
What you get
1–2 funded probes
With a kill criterion signed off before the probe starts, not written in after the fact.
A portfolio budget that holds
The number you set upfront stays the frame, not whoever argues hardest in the room.
The parked and blocked bets
Including the trigger that would bring a parked bet back into consideration.
Engagement model
What it costs
Fixed prices, which you get in the qualification call. The cost of the attempt is separate from the portfolio budget you set and from this engagement fee.
Opportunity AI
Fixed priceThrough probe design for 1–2 bets; running the probes is not includedProbe
Fixed price per probeScoped per probe at bet drafting; billed in full whether it succeeds or failsPerspective
When this is not for you
Opportunity AI is not for an organisation with no appetite to pay to be wrong. A zero portfolio budget is a no-go, and the same holds for a standalone probe. If you need guaranteed outcomes upfront, this is not the right service.
Next step
What is it worth to find out you're wrong here?
That's the first question we ask, not the last.