
5/ Okrona's point, and mine: writing down who decides is not bureaucracy. It is permission not to answer. Who has the final call on a schema change in your organisation? buff.ly/DoUps5j

Helping CTOs & CIOs turn AI ambition into an operating model: feedback loops, governance, and execution across people, process, technology | CAIO @ Zühlke

5/ Okrona's point, and mine: writing down who decides is not bureaucracy. It is permission not to answer. Who has the final call on a schema change in your organisation? buff.ly/DoUps5j

4/ Okrona's fix: write down who decides on schema changes, on what ships, on the pager. And one rule: "an AI-generated critique is an input to the owner, not a claim on the owner's time."

3/ This matters for empowered teams. If anyone can contest their decisions for free, they spend the week answering. The autonomy is still on the slide. The time is gone.

2/ Csaba Okrona's explanation: the right to overrule an expert used to be rationed by friction. A credible counter-argument cost a week, so few people made one. That friction quietly allocated decision rights. Now it takes ninety seconds.

1/ A staff engineer spent three weeks on a migration design. A colleague from another domain ran it past Claude and posted nine objections. Two were fair, one was a real catch, six were generic. Answering all nine took him most of a day.

5/ An agent can act on its own. It cannot answer for itself. If one of your agents went wrong tonight, who answers for it, and could that person stop it within the hour? Survey: buff.ly/SObhuAb

4/ Machines cannot carry responsibility. The deployer chose the agent's permissions, systems and data. Autonomy reaches as far as your gates hold, and revocation is the last gate. If it takes a day, the agent has a day of autonomy nobody granted.

3/ Four UK regulators (CMA, FCA, ICO, Ofcom) answered this month: "An AI agent's autonomy does not diminish the responsibility of organisations that deploy it" to comply with the law. Inside the companies, 9.5% have a named person formally accountable for agent behavior.

2/ 52% had or suspected an agent incident in the past year. Of those, 27.7% held the AI vendor responsible and 24.7% held a named person in their own company formally responsible. 17.3% recorded it and changed nothing.

1/ 97% say they are ready for an agent incident. 6% could stop the agent within an hour. 46% would need more than a day to revoke its access to all systems. That is from an October survey of 900 tech leaders in the UK and US by Gravitee, which sells agent governance tools.

5/ Be predictable about when you decide. Stop promising what will work. How many dates on your roadmap were set before anyone tested the idea? buff.ly/oyKmlnp

4/ My view: leaders are right to want both. They ask for predictability in the wrong place. The rhythm can be predictable: when we plan, review and decide. The content cannot. Nobody knows in January which idea will work in June.

3/ He names SAFe: "It is a model that very clearly is optimized for predictability, and makes no serious attempt at achieving outcomes." Ask leaders directly and they say they need both.

2/ Marty Cagan wrote about outcomes over output for more than a decade. Now he says: "I was framing this all wrong." You cannot get outcomes without output, so that is no trade-off. The real one, he writes, is outcomes versus predictability.

1/ Nobody argues for output over outcomes. Companies still ship output, because what they really want is predictability.

5/ Blast radius decides where the human stays. The factory grows one class of work at a time. Which class would you let ship without a person first? buff.ly/TcjjNOS