Stratogenic Blog
Practical notes on strategy, evidence, AI deployment and the decisions that have to survive contact with real operating conditions.
Pinned
2026
EBL is a seven-part learning loop for bringing ambition into contact with reality without surrendering either hope or the right to revise: notice, understand, decide, act, observe, review, adapt.
A provider update can alter behaviour without changing the workflow’s name. Keep a small regression set, decision owner and rollback boundary attached to the operating system.
A signed order confirms a commercial decision, not the moment a customer’s work improves. Define first value early enough for sales, delivery and the customer to challenge it.
A contract notice, appointment or expansion story can justify a better question. It cannot tell you that an account has the problem, budget or appetite your offer requires.
When work repeatedly slips, the honest question is not always how to try harder. Sometimes the task, timing or working conditions need to change before progress can return.
A public work address may be usable in one B2B context and unlawful or unwise in another. Separate recipient type, personal data, message purpose and objections before sending.
Making Tax Digital changes the weekly evidence trail before it changes the quarterly submission. The useful preparation is to test how records survive ordinary working life.
Identity verification is no longer a distant Companies House reform. Founders need a small, owned record of who must verify, when and what could block the next filing.
The AI Act’s August 2026 enforcement point matters, but a calendar date cannot classify your system, name its owner or decide what evidence makes deployment proportionate.
A successful AI pilot proves that a particular system handled particular tests under particular conditions. Deployment needs a separate decision about risk, ownership, monitoring and what happens when the system is wrong.
A business plan becomes useful when a reader can see what is known, what has been assumed, what would change the numbers and which action will produce better evidence next.
Useful prospect research does not manufacture familiarity. It turns public evidence into a challengeable account hypothesis, names what remains unknown and earns a proportionate next conversation.
A signed order and a tidy internal handover can still leave the customer outcome unowned. The remedy is to reconstruct the promise, separate launch from value and assign the evidence needed for first value.