Stop Fighting With AI. Start Co-Consulting With It Instead

Last week I showed you how I used AI to extract my own framework for using AI.

I also just published a YouTube video walking through the process.

When I did the video, I decided to use Claude instead of ChatGPT

Same raw material (the 2-hour transcript of me helping a consultant turn 80 pages of notes into a client-ready report).

Claude gave me cleaner step names, tighter logic, and a framework that's easier to remember.

Here's what Claude pulled out of that transcript. I’m sharing it here so you can apply it to your own interactions with AI (basic principles apply, even if you’re not writing a report).

The Co-Consulting Report Framework

Five steps. Each one matters.

1. Prepare – Set-up Your Workspace
Set up a dedicated AI project. Upload all your source documents (research, data, background files) to the project. Write custom instructions that define the AI's role, audience, and output style.

But DON'T upload your draft or working notes yet. Keep those separate. This causes AI to treat your rough notes as authoritative source material)

2. Plan – Build the structure before you write anything
Now upload your draft notes to a conversation. Ask the AI to suggest an outline based on your content. Don't mandate the structure—give examples and let it suggest what fits. Review it. Adjust it. Then copy the final outline to a separate document you can reference throughout.

This outline becomes your map. Everything you write follows it.

3. Execute – Draft one section at a time with validation
Work through sections sequentially. For each section, tell the AI: "Draft [section name]. Use the draft report and project documents as your source. Ground the analysis in the actual data." And paste in the full outline for that section to remind the AI what you had agreed it should include.

Add specific facts the AI doesn't have (dates, names, details). Give it tone and length guidance. And when the AI drifts or invents something, STOP. Ask "where are you getting this?" and force it back to your sources.

Most people accept bad output and fix it later. That's wasted energy and time.

4. Iterate – Refine complex sections through options
For recommendations or strategic sections, don't accept the first structure. Ask for options. Try different ways to organize the same material.

Example: AI gives you sequential options (Option 1, Option 2, Option 3). Ask it to reorganize thematically instead (by cost, by risk, by impact). Compare both. Pick what serves your audience better.

5. Complete – Write the summary LAST
Generate the conclusion. Then—and only then—generate the executive summary based on the actual report content you've validated.

Most people write the summary first. Then they change the report. Then the summary is wrong. Then they rewrite it. Then they change something else.

Backwards.

Write the substance. Validate it. Then summarize.

The difference between the ChatGPT version and the Claude version wasn't huge.

But it was enough.

ChatGPT gave me: Define, Structure, Anchor, Draft, Synthesize.

Claude gave me: Prepare, Plan, Execute, Iterate, Complete.

And that's the thing about extracting your own frameworks:

The first version is rarely the final version.

You extract it. You use it. You refine it.

Sometimes you run it through a different tool and see it from a different angle.

This framework came from one transcript. But I've used this process on dozens of projects over the past few years. I just never articulated it clearly until now.

Now I can teach it. Now I can refine it. Now I can turn it into products.

When I reached out to the owner of a consulting company I’ve worked with to see if he’d like me to teach it to his consultants, he called it a 'brilliant offer’.

And now you can steal it.

If you write reports, strategic docs, client deliverables, research papers, or anything longer than 5 pages—this framework will save you hours.

Not by having AI think for you.

By having AI organize YOUR thinking faster than you could type it yourself.

👉 Watch the YouTube video: How I Extracted This Framework (step-by-step walkthrough of the meta-analysis process)

👉 Last week's issue: How to Extract Your Own Framework Using AI

👉 Get the full framework: The AI Co-Consultant Framework (complete guide with examples, common mistakes, and implementation checklist)

Thanks for reading!

Nathan Rodgers

👋 Say hello on Substack and LinkedIn

Were you forwarded this email? Click here to subscribe.