AI strategy01
Before you buy another AI tool, map one day of work.
A practical way to find a useful AI starting point: follow one task, identify the friction, and define a result worth measuring.
The journal / Leadership / Note 06
Leadership · Field note 06Five questions a leader should ask before the rollout.
A launch announcement tells people that a tool is available. It does not necessarily tell them how their work is meant to change. That requires a more specific conversation.
What is supposed to become better?
Name the task and the intended improvement. “Help the team work smarter” is a broad aspiration. “Reduce the effort needed to prepare a checked internal project update” describes something people can recognize.
Be careful about what sits outside that goal. Faster drafting does not automatically mean better decisions, happier customers, or less work overall. If those outcomes matter, decide how they will be examined rather than treating them as a consequence of adoption.
Who is responsible for the finished work?
People need to know who may use the tool, who checks the output, and who approves a change to the workflow. These responsibilities should survive a busy week or a colleague’s absence.
A named owner should also keep the instructions current. If the source material, the tool, or the task changes, someone needs to decide whether the existing review routine still makes sense. Responsibility becomes vague when everyone assumes it belongs to someone else.
What does a good result look like?
Create a small set of examples that show both an acceptable result and a plausible failure. Explain the difference. A missing condition, an invented deadline, or a suggestion presented as a decision can matter more than a visible spelling error.
Review should be connected to the work’s purpose. A summary may need accurate attribution and clear open questions. A draft announcement may need a checked action list. Generic approval criteria can miss the details that make a particular task consequential.
What happens when it goes wrong?
Give people a clear way to question an output, report a recurring problem, and return to the established process. The response should be practical enough to use under pressure.
NIST’s AI Risk Management Framework is a broader reference for thinking about AI risk. For a specific rollout, translate the relevant concerns into ordinary working instructions: who to contact, what to pause, which material to retain, and who decides what happens next.
What evidence would justify doing more?
Define what you need to see before adding users, tasks, or access to information. A positive reaction to a demonstration is encouraging, but it is different from evidence that a workflow produces reliable results over repeated use.
Ask for a short account of the results and the limitations. Then decide whether to expand, revise, or stop. A leader’s job includes protecting the quality of the decision, even when the proposed technology is exciting. Clear questions give the team permission to be honest about what is working.
One thing to take away
Make the intended change, ownership, review standard, fallback and expansion criteria clear before the announcement.
Further reading
NIST — AI Risk Management Framework ↗
A reference for organizations considering how to manage AI-related risk.
Pasqualino Mazza
Entrepreneur, global AI consultant, and content creator. Writing about AI, business, and leadership from a Toronto perspective.