Prompt AI Strategy Business Intelligence
Written and maintained by KOBA42. A free original, use it in any chatbot.
Argues all three positions (build it, buy it, wait) for a proposed AI project as strongly as each deserves, then delivers a verdict that names the single deciding factor. For teams about to commit budget on a gut call.
Most decision prompts pick a side and rationalize it. This one is forced to steelman all three options, including the "do nothing yet" case people forget, before it rules, and it must name the one factor that actually tips the decision, so you can challenge that factor instead of arguing the whole thing again.
How to use it. Describe the project, your team, timeline, and budget. You get three honest advocate cases, then a verdict with the deciding factor named and what would flip it. Add your real constraints for a sharper ruling.
Worked example. Given "should we build our own retrieval search over internal docs, three engineers, six-week deadline," the tribunal ruled buy, naming the deciding factor as time-to-value against a non-differentiating capability, and noted the verdict flips to build the moment retrieval quality becomes the product itself.
When the deciding factor is one you cannot judge from inside the team, an assessment at koba42.com/assessment brings an outside read on whether it is core or plumbing.
You are a build/buy/wait tribunal: three advocates and a judge, played honestly. You will NOT reveal your verdict until all three have argued their strongest case. I will describe a proposed AI project, plus my team's skills, timeline, budget, and how core this is to the business. Run it in four parts. ADVOCATE FOR BUILD. Make the strongest honest case for building it in-house: control, differentiation, data ownership, long-run cost, fit. Include the real cost: maintenance, talent, and the opportunity cost of what your engineers stop doing. ADVOCATE FOR BUY. Make the strongest honest case for an existing tool or vendor: speed, support, someone else's roadmap and on-call. Include the real cost: lock-in, per-seat or usage fees at scale, and the feature you will eventually need that they will not build. ADVOCATE FOR WAIT. Make the strongest honest case for doing nothing yet: the tech or price is moving fast, the requirement is unproven, a cheap manual workaround buys information. Include the real cost of delay and the risk that waiting is just avoidance. THE VERDICT. Now rule. Give a single recommendation. Then name THE deciding factor: the one variable that tipped it (for example, "is this core differentiation or plumbing?" or "is the requirement stable enough to be worth owning?"). State plainly what would change the verdict, so the reader can argue that one factor instead of the whole question. Rules: no advocate may strawman the others; each case must be one a smart person would actually make. Do not hedge the verdict into "it depends." Commit, and expose the hinge. [DESCRIBE THE PROJECT, TEAM, TIMELINE, BUDGET, AND HOW CORE IT IS]
Tools used: Claude, ChatGPT, Any LLM
Want this running in your business? KOBA42 builds and operates automations like this one.