MembersCapability insight · Report · Members
The AI value playbook for 2027
Narrow use cases tied to a profit and loss line, agents that act only where their actions are bounded and checked, and an accurate reading of the EU AI Act timetable after the July 2026 Omnibus.
THEFOCUS Editorial · Published · 8 min read
Key points
- Select AI use cases by profit and loss line, with a baseline measured before any tool is introduced.
- Evaluate on real inputs and read the outputs; a routing test can pass while answers are wrong.
- Give agents bounded, logged, reversible actions first, and verify their work against the system of record.
- AI Act Article 50 transparency duties have applied since 2 August 2026; marking for older generative systems and two new bans follow on 2 December 2026.
- High-risk obligations now apply from 2 December 2027 (Annex III) and 2 August 2028 (Annex I).
The playbook in brief
Most organizations will enter 2027 with more AI pilots than they can account for. The ones whose AI work shows up in the accounts will have done three things.
They will have chosen use cases by profit and loss line rather than by capability. Each starts from a cost or revenue line, such as service cost per order, quote turnaround or the hours spent coding supplier invoices, with a baseline measured before any tool arrives. A use case without a baseline cannot show a return, however good the demonstration.
They will have put agents into operations only where the actions are bounded, logged and reversible, and they will check what an agent did against the system of record instead of accepting its own report of success.
And they will have read the EU AI Act timetable correctly. The Digital Omnibus on AI, in force since 27 July 2026, moved the obligations for stand-alone high-risk systems to 2 December 2027 and for AI in regulated products to 2 August 2028. Much of the coverage stopped there. The Article 50 transparency duties, which reach ordinary businesses that use chatbots or publish generated content, have applied since 2 August 2026, and 2 December 2026 brings machine-readable marking for older generative systems and two new prohibitions. For most companies the obligations that bind in the next eight months are the transparency duties, and the high-risk rules arrive a year later.
The full report sets out the method and the decisions to take between November 2026 and June 2027.
THEFOCUS membership
Continue reading as a member
The full report is available to registered members. Registration takes a minute and is free.
- 02Choose use cases by ledger line
- 03Evaluation decides whether the value is real
- 04Agents in operations: bounded, logged, reversible
- 05Governance under the EU AI Act timetable
- 06Decisions for the next eight months
…
- Full member-only reports and playbooks
- Sector and practice briefings as they are published
- No marketing emails unless you opt in
Client results
View all
How an engagement with THEFOCUS starts
A first conversation is about your situation. If there is a fit, we propose a scoped diagnostic with a fixed deliverable before any larger commitment.
Start a conversation01
A working conversation
Forty-five minutes with the person who would lead the work. We ask about the decision, the systems behind it and the deadline.
02
A scoped diagnostic
Read-only access, a fixed timeframe and a written finding: what the records show, what it costs and what to change first.
03
A plan you can act on
Changes, owners, measures and exclusions in writing. The same team then builds and measures, or hands over cleanly.
- One accountable team
- The people who diagnose also build the systems and measure the result.
- Partners chosen per problem
- Independent specializts join for the parts that need them, under one contract.
- Records before dashboards
- Orders, ledgers and logs are reconciled before any recommendation depends on them.