AI COLLABORATION
Book a conversation
Don’t let AI slop become your standard.
Decide what the work needs from people, then choose how AI can help. You may want to explore more possibilities, strengthen existing work, or reduce routine production. Each choice affects originality, audience relevance, review effort, skill development, and trust.
Across teams, agencies, and tools, those choices can become inconsistent. This guide helps you make them explicit and build the guidance, review, and working practices needed to support them.
Write down the business goal, what the audience should understand or do, and why it would matter to them. Consider their relationship with your organization and expectations of the communication. Give that direction to both the people and AI involved.
These modes are choices for each stage of the work, not steps to complete in order.
Increasing delegation to AI, not increasing quality or maturity.
| Consider | Human creation | Thinking partner | Editor | Co-creator | Delegate |
|---|---|---|---|---|---|
| Why choose it | The act of creating matters. Choose it for personal expression, learning, or work grounded in lived experience. | Explore before deciding. Consider more possibilities and perspectives, or challenge a familiar approach. | Strengthen work worth preserving. Improve clarity and execution while keeping the underlying direction and distinctive choices. | Develop your ideas further. Turn substantial human thinking and expertise into finished work through iterative collaboration. | Hand over a bounded task. Choose it when the standard is explainable and people can reliably evaluate the result. |
| How the work is shared | People develop and make the work. AI is absent from the working session. | AI explores; people choose and create. AI proposes possibilities, questions, and counterarguments. People determine the direction and make the work. | People supply the work and approve changes. AI critiques, checks, or refines within an agreed scope. | People direct, select, and revise. AI helps develop and produce work from substantial human ideas, experience, references, and direction. | People define the outcome and boundaries. AI develops much of the approach or output, with human review at agreed points. |
| Tradeoff | Time and effort are part of the work. They may be part of its purpose, not something to eliminate. | Suggestions can steer your thinking. Familiar patterns and reinforced assumptions can narrow the exploration. | Polish can change meaning. Personality and distinctive choices can disappear while the result looks more finished. | Active direction takes time. Polished output can obscure weak substance, so people still need to select, question, and revise. | Checking and correction can erase production savings. Dependable inputs, clear instructions, and reliable human evaluation remain necessary. |
Human intention and accountability remain throughout.
My principle is Human → AI → Human: work begins with human intention and returns to human judgment. It does not prescribe one fixed workflow.
Give people and their AI tools the relevant goals, approved sources, audience knowledge, standards, and examples. Explain what must be preserved and what is out of bounds. Across a team or agency, make the guidance accessible in the tools and working practices they actually use, with someone responsible for keeping it current.
This is the main working tool. Complete it with the people who create, review, and distribute the work. Record how the purpose, AI role, and source material connect to decisions and handoffs. Agree what must be checked before the work moves on.
| Question | What to record |
|---|---|
| What should the communication accomplish? | Business goal, audience, relationship, and desired response. |
| What role does AI play? | Research, exploration, creation, adaptation, evaluation, or distribution. |
| What informs the work? | Approved sources, current facts, context, gaps, and source owners. |
| Which standards apply? | Brand, quality, audience, legal, commercial, and confidentiality requirements. |
| What can the people and systems access? | Relevant guidance, information permissions, and what may enter each tool. |
| Who makes which decisions? | Creator, reviewer, approver, escalation owner, and authority to change system instructions. |
| Where does the work go next? | Partners, channels, platforms, other AI systems, and changes after approval. |
| What evidence stays with the work? | Sources, permissions, decisions, approvals, and who checks the released version. |
| How do corrections improve future work? | Feedback records, testing, approval, and versioning. |
| How will we measure improvement? | A baseline and a few measures tied to the workflow’s purpose. |
The access, evidence, approval, and release questions connect this worksheet to compliance. See AI marketing rules for a deeper guide. Governance also means deciding who can change the system’s standards and how those changes are tested and approved.
Once the workflow is in use, capture what needs to change and why.
Decide whether the feedback is a durable rule, a campaign preference, an exception, or a response to new information.
AI can identify patterns and propose changes. Designated human owners decide what becomes authoritative. Record the scope of each rule so a campaign preference does not quietly become a company-wide requirement.
Check whether the change improves the outcome you intended, not just the volume produced. See measurement guidance.
Use the deeper guidance below where the worksheet reveals a gap. The sections follow the same process, with additional help on roles and measurement.
Support these questions with practical examples and a named escalation owner:
Teach the framework through real work so people can recognize uncertainty and know whom to ask.
Begin with the outcome and the audience. What do we want people to preserve, experience, develop, or accomplish through this work? The answer might be personal expression, learning, exploration, communication, or efficient adaptation.
Choose AI’s role for each stage. You might develop an idea without AI, explore alternatives with it, use it to produce variations, and return to human editing and approval.
Choose based on intention, source material, audience expectations, and your ability to evaluate the result.
What happens: AI is absent from the working session. People develop and make the work directly.
Choose it when: The act of creating is part of the goal: discovering your own thinking, practicing a skill, drawing from observation, or expressing something personally.
Example: Sketch a campaign concept from your own experience before looking at generated alternatives.
Tradeoff: More human effort, with room for independent thought and discovery.
What happens: AI helps explore questions, possibilities, assumptions, and counterarguments. People determine the direction and make the work.
Choose it when: You want to cover more ground, consider another perspective, or challenge a familiar approach.
Example: Explore different audience interpretations of a concept, then have the team develop its own creative response.
Tradeoff: Suggested ideas can anchor the discussion or reinforce assumptions. Ask for credible alternatives and bring in human perspectives.
What happens: People supply existing work. AI critiques, checks, or helps refine it within an agreed scope.
Choose it when: You want to improve clarity, consistency, accessibility, or execution while preserving the underlying direction.
Example: Review a presentation for unsupported claims, unclear hierarchy, or departures from brand standards.
Tradeoff: The system may flatten distinctive choices or alter meaning while polishing. Specify what it may change and what must be preserved.
What happens: People supply substantial ideas, experience, references, and direction. AI helps develop and produce the work through an iterative collaboration.
Choose it when: You have a clear point of view or creative intention and want help exploring expressions, combining material, or producing alternatives.
Example: Develop campaign copy, visual directions, or a presentation from interviews, approved sources, and a human creative brief.
Tradeoff: Directing, comparing, and revising takes time. Set a stopping point and judge the work against the goal.
What happens: People assign an outcome and boundaries. AI develops much of the approach or output, with human review at defined points.
Choose it when: The task is bounded, the inputs are dependable, and you can tell whether the result meets the standard.
Example: Produce initial channel variations from an approved campaign, within defined claims, formats, and review requirements.
Tradeoff: Delegation can save production effort while increasing the importance of clear instructions and evaluation. With a thin brief, AI supplies more of the substance and assumptions itself.
Complete this before you begin:
We are using AI as [mode] to help achieve [outcome] for [audience]. People will contribute [knowledge, creative direction, or work]. We will preserve [what matters]. [Owner] will evaluate the result against [criteria] before [next step].
Illustrative example: We are using AI as a delegate to adapt an approved campaign for three channels. People supply the strategy, approved claims, assets, and audience context. The brand reviewer checks meaning and distinctiveness; the channel owner checks format and release settings. A new claim returns for specialist approval.
A mode is a choice for the work in front of you. Change it when the purpose, context, or stakes change.
Make three things usable in the workflow: documented guidance, expert judgment, and the production process.
Documented guidance includes business goals, values, brand guidelines, audience knowledge, and relevant requirements. Expert judgment includes what experienced people mean when they say something feels off, misses the audience, or fails to sound like the company. Draw that out through examples, specific edits, and the reasons behind them.
Map the process too: where work originates, which tools touch it, who approves it, and what happens after approval. Identify where information enters another AI system or where context and control can be lost.
Then configure the enterprise tools around those decisions: reference materials, instructions, guided intake, connected sources, review checks, and approval workflows. Assign people to keep them useful and current.
The perspective behind this guide: During 15 years at HBO, I helped introduce and scale digital marketing, social media, and emerging technology for a company with exceptionally high creative standards. New platforms and forms of communication brought new possibilities, large audiences, and real brand risk. That experience informs how I help organizations use AI while preserving judgment, creative ambition, and audience trust.
In one enterprise engagement, setting up an agent with brand guidelines was straightforward. The harder work began when the output still felt generic or missed the goal of the communication and the role of the audience.
We reviewed outputs with experts, recorded their edits and reasoning, and translated general feedback into specific, repeatable guidance.
We also created a guided experience to help employees clarify what they were trying to accomplish. The work improved the guidance inside the system and how people expressed their goals.
Consider visibility, information sensitivity, audience expectations, legal and commercial obligations, novelty, and the consequences of getting something wrong.
Assign reviewers and decision rights for your organization. For example:
| Situation | What deserves attention |
|---|---|
| Adapting an approved description for a familiar channel | Check that meaning, claims, facts, and required language survive the adaptation. |
| A personal message from a senior executive | Review the relationship, voice, intention, and whether the message represents what that person means. |
| A new public claim or sensitive announcement | Involve the relevant business and subject experts, with legal or other specialist review where required. |
| Content changed by an advertising platform after approval | Understand what can change, what controls exist, and how the delivered version will be checked. |
Uncertainty itself should trigger escalation. A team member should know whom to ask and have the time and authority to pause.
Test AI review against expert examples. Check both missed problems and unnecessary flags, then refine the review criteria.
Plan for the work upstream. People need time to improve sources of truth, interpret inputs, test systems, evaluate recommendations, and refine standards. Put those responsibilities into roles, training, and workload planning.
Keep creative ambition visible. Fewer preventable corrections should create more room for creativity, empathy, strategic thinking, and differentiation.
Ask the team: What work can we stop doing, what new stewardship does the system require, and who has the capacity and authority to do it?
| If the intention is to improve | Look for |
|---|---|
| Efficiency | Less rework, fewer review rounds, shorter time to an approved result. |
| Baseline quality | Fewer repeated errors, better factual support, stronger brand alignment. |
| Creative effectiveness | More distinctive ideas, clearer audience relevance, stronger expert and audience response. |
| Risk visibility | Appropriate issues caught and escalated, missed issues, and unnecessary flags. |
| Learning | Useful lessons captured, tested, and applied to later work. |
| Business contribution | The relevant change in understanding, perception, engagement, action, or another defined outcome. |
Research and experimentation need objectives too. Measure useful learning alongside production and performance.
Knowing which role AI should play is a starting point. Making that choice work across a team also requires usable guidance, the right configuration, and people who know what to review.
I help translate your intentions, expert knowledge, and standards into guidance people and AI can use. Together, we can configure your enterprise tools and put that guidance into practice through clear handoffs, review, and feedback.
My approach is to meet your team where it is and focus on a change that can benefit the work now.
Let’s look at one content or creative workflow, identify what needs to improve, and determine what to build together.