Seven Hats — Build Your AI Marketing Assistant: a client's view
Purpose: Assess whether a prospective nontechnical small-business client would understand the offer, trust it, and complete the exercise independently.
Document summary
This workshop deck teaches solopreneurs to improve AI-generated marketing by supplying their writing style, audience, business details and point of view. Participants first generate a generic LinkedIn post, build a reusable marketing context file, then compare a second post created with that context. The approach keeps the business owner responsible for ideas and final editing. The closing exercise directs participants to the presenter's Online Signal Report and AI Visibility Score to identify a marketing priority.
Blue — Overview
- The clearest deliverable is a reusable marketing context file plus a revised LinkedIn post (pp. 15–22). The broad title could suggest an assistant that manages marketing; this session primarily builds context for writing. State that deliverable at the start.
- This is clearly designed for a live session: repeated requests to comment and references to demonstrations rely on facilitation (pp. 4, 6, 10, 12, 23, 28). A client receiving only the PDF needs a companion worksheet and setup guide.
White — Evidence
- Source claim: the 10–80–10 split is presented as a goal (p. 11), not supported by time measurements. No before-and-after output, case study or conversion evidence is included to demonstrate the business outcome of bringing in clients (pp. 7, 22). Treat the split as an aspiration, not a measured saving.
- The prompt explicitly prohibits invention and requires missing information to be marked as a gap (p. 20). This is a useful safeguard; the client still needs to inspect the generated claims and voice rules.
- The AI Visibility Score is introduced without an explained scoring method, tested platforms, repeatability or connection to enquiries (pp. 25–27). Its predictive value is not verified. Presenter credentials on p. 2 are source claims and were not independently checked.
- Verified narrow setup point: OpenAI documentation distinguishes project files from project instructions and makes Projects available across free and paid plans. Page 21's ChatGPT Pro label and instruction to add the file as instructions need clearer, platform-specific explanation; they can imply unnecessary subscription requirements. This does not establish Claude requirements.
Red — Response
- Interpretation: the familiar frustrations about generic tone and hours of editing may make a client feel understood (p. 5). Permission to give rough answers reduces pressure (pp. 17–18).
- Interpretation: named frameworks, hot takes and proof stories can feel demanding to a beginner who has expertise but no packaged methodology (p. 19). An ordinary service-business example would make this more approachable.
- Interpretation: the opening question about what AI says about you creates curiosity, but can also generate worry (p. 1). Moving into a score and the presenter's services may feel like a sales transition unless the report's purpose and limits are made clear (pp. 26–29).
Yellow — Benefits
- The starter kit asks for concrete inputs clients can supply: one writing sample, audience needs, an offer and a point of view (pp. 15–19). Reusing those inputs could reduce repeated briefing and make later drafts more consistent.
- The same-prompt comparison gives participants a practical way to notice changes (pp. 3, 22). It becomes more useful with explicit criteria: recognisable voice, audience relevance, factual accuracy and editing time.
- Keeping the owner responsible for the idea and human review supports expertise-led content (pp. 9–11). These benefits depend on accurate source material and active review; they are plausible mechanisms, not demonstrated client-acquisition results.
Black — Caution
- Highest-priority friction: saving Markdown and placing it correctly in a project are compressed into a few lines (pp. 20–21). A novice can complete the thinking yet fail at the technical handoff. Provide separate illustrated steps for each platform, including a quick check that the context is being used.
- Expectation gap: capabilities, connections and cadence appear in the wider assistant model, while today's exercise covers context (p. 13). Clients could expect scheduling or integrations that the exercise does not configure. Define what is completed today and what remains future work.
- Client transcripts and real stories are suggested as source material without practical guidance on removing identifying or confidential details (pp. 9, 17, 19). Add a short anonymisation step and a reminder to use material the client is entitled to share. This is a workflow concern, not a legal finding.
- The deck moves from content quality to AI visibility without demonstrating the causal link (pp. 25–27). A better-sounding post, a higher score and a paying client are different outcomes. Set separate measures for each.
Green — Possibilities
- Proposal: add one worked example using a familiar small business: original prompt, first draft, completed context, improved draft and human corrections. Explain what changed and why, without implying a guaranteed result.
- Proposal: add a one-page start-to-finish checklist covering prerequisites, realistic completion time established through testing, platform-specific setup, a downloadable context template and a fallback for clients with no writing samples.
- Proposal: add an approval checklist before publishing: accurate facts, no invented results or testimonials, recognisable voice, one audience problem and one clear action. Track editing minutes and relevant enquiries separately from impressions.
- Proposal: explain the visibility report's inputs, scoring basis, limitations, data handling and any follow-up offer before asking clients to enter details. Label it as a diagnostic aid unless evidence supports stronger claims.
Purple — Connections
- Interpretation: the context file may be the most durable asset. The instruction to make it understandable to another AI (p. 20) could support consistent briefing across tools or collaborators. Its value depends on keeping offers, proof and audience assumptions current.
- Interpretation: clearer content can reveal weaknesses elsewhere in the client journey. The business questions ask what clients should do next (p. 18), but results also depend on the offer, landing page and response process. Better drafting alone cannot resolve those dependencies.
- Possible second-order effect: rigidly following one voice sample or repeatedly pursuing provocative hot takes could narrow the client's voice or damage trust (pp. 16, 19). Use several representative samples over time and permit disagreement only where the owner has a defensible view.
Synthesis
Alternative reading
The strongest alternative reading is that this is a slide deck supported by live demonstrations, discussion and additional resources. Those may resolve the setup and evidence gaps. A session recording, companion workbook, complete example and explanation of the visibility score would materially improve the assessment; their absence from this PDF does not establish their absence from the training.
Conclusion
A practical, approachable workshop foundation for improving AI-assisted writing. As a standalone client handout it needs clearer setup, a complete worked example and tighter expectations about marketing outcomes. A client's likely response is: I understand why context matters, but show me exactly how to finish and how I will know it helped my business.
Confidence and unknowns
High confidence in the content and standalone-usability findings, based on all 29 pages. Moderate confidence in likely client reactions because no client testing or live session was observed. Low confidence in any claim about lead generation or the visibility score because supporting outcome evidence was not supplied.
Next action
Proposal: test the PDF with one nontechnical target client, without live guidance, and use the points where they stop to create a one-page setup companion and worked example.
References
- Source PDF: Build Your AI Marketing Assistant, all 29 pages; page numbers refer to PDF order.
- OpenAI Help Center — Projects in ChatGPT; narrow external check of project setup and availability