Consulting for creative, marketing and communication teams
AI Use Audit for Teams
AI can make work faster while also increasing the risk of weak verification, poor decisions and unclear ownership. I help creative, marketing and communication teams examine how they actually use AI and decide which working principles need to be clarified.
Who it is for
When an AI use audit is useful
- The team uses ChatGPT, Copilot, Claude or other models without shared rules.
- AI-generated or AI-assisted material reaches clients or external audiences.
- It is unclear who verifies an output and owns the final decision.
- People share sensitive, confidential or poorly organized material with models.
- AI supports the reading of briefs, messages, recommendations or decisions.
- Fluent model answers are accepted without checking their sources or assumptions.
- Different people use AI in completely different ways across the same process.
What we examine
The audit follows a real workflow from input to final material. It does not depend on one model provider or a single tool.
- 01
Where AI enters the process
Tasks, handoffs, decisions and deliverables in which a model already plays a role.
- 02
Data, inputs and privacy
What people provide to models, how sensitive that material is and whether boundaries are understood.
- 03
Sources and verification
Where claims, summaries and generated material come from, and who checks them before use.
- 04
Decision and output ownership
Who approves the result, who can question it and who remains accountable for consequences.
- 05
Trust in model outputs
Where fluency, speed or confidence may be mistaken for reliability.
- 06
Communication, interpretation and sycophancy
How a model frames messages, reinforces assumptions or turns one account into an apparent verdict.
- 07
Division of work and human control
Which tasks AI can support, which require human judgement and whether people can inspect, challenge and revise the result.
- 08
Rules the team can actually use
Simple principles, review points and ownership rules that fit everyday work.
Process
How the audit works
The exact scope depends on team size and the number of workflows being examined. The process stays focused on a small set of real cases rather than an abstract catalogue of tools.
- 01Opening conversationGoals, team context and the questions that need a clearer answer.
- 02Case selectionA few real workflows or use cases that matter to the team.
- 03Use analysisInputs, model outputs, checks, handoffs and decisions.
- 04Risk mapVerification gaps and places where responsibility is unclear.
- 05RecommendationsPrioritised principles and practical changes for the selected workflows.
- 06Review sessionA meeting or workshop to discuss the findings with the team.
Deliverables
What the team receives
- A map of how AI is used in the selected workflows.
- A prioritised list of the most important risks.
- Clear points where sources, assumptions or outputs require verification.
- A proposed division of work between people and AI, including decision and review responsibility.
- Recommendations for working with material, sources and communication.
- A written summary that the team can return to.
- A review meeting or workshop for the team.
Boundaries
What the audit does not do
- It is not a legal audit.
- It is not a cybersecurity audit.
- It does not replace a data protection specialist.
- It is not a psychological assessment of employees.
- It does not transfer decisions or responsibility to AI.
- It does not promise to eliminate every error.
First step
Describe the situation
Briefly describe:
- what kind of team it is,
- where you already use AI,
- what creates the most uncertainty.
Based on that context, I will reply whether the AI Use Audit is a useful next step and what scope would be proportionate.