How to Build an AI Advisory Panel to Test a Business Decision

Key takeaways

• A panel works best when it votes on clear options, not an open question
• Give each adviser one question to own, so they do not all say the same thing
• Run it in three rounds: opening positions, challenges, final vote
• Ask for assumptions and risks alongside the recommendation
• AI advisers are simulations; you still own the decision

This guide supports our main article, 3 Claude Code AI Workflows Nate Herk Uses Every Day (and How to Copy Them). In it, Nate Herk shows the Roast Room, a Claude Code panel of AI business advisers that debated Matt Wolfe’s newsletter strategy live. Here we answer one practical question: how can you build a simpler AI advisory panel to test your own business decision, using any AI chatbot?

What is an AI advisory panel?

It is a structured conversation where an AI plays several advisers, each looking at your decision from a different angle. Instead of asking “should I do this?” and getting one agreeable answer, you get competing views, a debate and a vote.

This matters because AI chatbots tend to agree with whoever is asking. Forcing several viewpoints to disagree is one of the simplest ways to get past that. It is the same idea we cover in How to Get Honest Answers From an AI Chatbot, scaled up into a panel.

Step 1: How do you frame the decision?

Write one paragraph that a stranger could understand. Include what you are deciding, why now, your budget and time, how you make money today and what success looks like.

For example: “I run a small online course business making about 70% of revenue from email. I am deciding how to spend the next three months of marketing time. I have about 10 hours a week and £1,500.”

Step 2: Why must you give the panel options?

Because a panel that votes on options gives you a decision, while an open question gives you a lecture. In the video, Nate’s one rule before running the Roast Room was that it needed options. If you only have one idea, ask the AI to suggest three alternatives first, then add “do nothing” as a baseline.

Step 3: How do you choose your advisers?

Pick four to six advisers who would naturally disagree, and give each one a single question to own. Nate gives every adviser one question; his AI Alex Hormozi, for instance, judges whether people will pay and how you close the sale. You can use well-known figures or plain roles. Roles are often safer and clearer.

Step 4: How do you brief each adviser?

Give each adviser a short profile: their question, how they think, what they value and what they would never recommend. Nate’s versions draw on large research files, but three to five lines each is enough to start. The more distinct the profiles, the better the debate.

• Their question: the one thing they judge every option on
• Their lens: for example “cash first, cautious about debt”
• Their style: blunt, data-led, customer stories, and so on
• Their red lines: what they will always push back on

Step 5: How do you run the debate and vote?

Run three rounds, in order, in the same chat.

1. Opening positions: each adviser picks an option and gives their reasons in three bullet points
2. Challenges: each adviser picks one other adviser they disagree with and challenges them; the other replies. Run these one at a time so later advisers can see what was said, as Nate does
3. Final vote: each adviser votes, says whether they changed their mind and why, and scores every option out of 10

Finish by asking for a one-page summary: the winning option, the vote count, the key assumptions, the biggest risks and the first three actions.

Step 6: How do you check the verdict?

Check the reasoning, not just the winner. Look at what assumptions the panel made. In the video, the panel estimated about $78,000 a year at risk for Matt, but said itself that the figure rested on assumed numbers. Yours will do the same.

• Underline every number and check it against your own data
• Ask “what would make the losing option the right one?”
• Run the panel a second time with the options in a different order; if the result flips, the case is close
• Sleep on it, then decide yourself

Can I copy a ready-made prompt?

Yes. Paste this into ChatGPT, Claude or Gemini and fill in the brackets:

What are the limits of an AI panel?

An AI panel is only as good as the context you give it, and it cannot see your real numbers or customers. It is useful for spotting blind spots and comparing options quickly. It is not a replacement for real customer feedback, an accountant or a solicitor.

Nate also found that running each adviser as a fully separate agent team cost more without much extra value, so a single well-structured chat is a sensible place to start.

Frequently asked questions

The bottom line

An AI advisory panel turns one agreeable chatbot answer into a proper debate. Frame the decision, give real options, assign each adviser one question, run three rounds and check the assumptions. Your next step: try it on one real decision this week.

Related posts

• 3 Claude Code AI Workflows Nate Herk Uses Every Day (and How to Copy Them)
• How to Get Honest Answers From an AI Chatbot
• How to Choose the Next Business Workflow to Automate With AI

Garry Knight

I'm Garry Knight, the person behind Prodify Digital. I write about email list building, email marketing, SEO, AI search and the tools that connect them. My aim is to make online marketing easier to understand, so creators and small business owners can make informed decisions about building an audience and keeping people engaged. Here you'll find straightforward guides and product reviews that explain what something does, where it fits and which limitations matter. The focus is on clear explanations and useful next steps—not hype, shortcuts or promises of easy earnings.

See Full Bio