How do I build an AI board of advisors that tells me the truth about my work?

Five personas, not one chatbot, because five of them argue We built five AI advisor personas instead of one. They are modeled on the public ideas of MrBeast, Joe Rogan, Naval Ravikant, Seth Godin, and David Goggins. None of those people took part, and none of them saw any of this. The point of five is that they pull in different directions, so you get a disagreement instead of a chorus. One voice would have agreed with us and sent us on our way. Five voices make the gaps show.

Three of five voted for depth, and that vote became the decision We forced one tradeoff question with three fixed options. As the steps table shows, three said depth, one said the third option, and one said growth. Counting the votes turned five opinions into one decision, so the board said go for depth. Open questions get you essays. A forced vote gets you a call you can act on this week.

The 8 steps end in a dated 30-day change, not a feeling The steps table on this page has 8 rows, and the last one is the part most people skip. In the video, Nick commits to 30 days of changes: tighter structure, tighter intros, and cutting anything that doesn't earn its spot. Advice that never gets a date on it is just a nice conversation. A date turns a chat into a test you can pass or fail.

What are the 8 steps to build an AI board of advisors?

These are the steps as Nick lays them out in the video, in order, with the timestamp where each one starts. All 8 rows are here.

Rows copied from the video walkthrough. Times are seconds from the start.

What happens at each point in the video?

All 10 key moments from the video, in order, so you can jump to the part you need.

10 moments, in the order they happen.

How do I build an AI board of advisors?

You build it in 8 steps, and the whole thing rests on step 3: feeding the board your real data instead of your pitch about your work. We picked five voices, gave them permission to disagree, wired in real transcripts and real numbers, asked wide, forced a vote, asked the hard private questions, argued back, then set a date.

The steps table on this page holds all 8 rows in order. Step 1 starts at 19 seconds into the video. Step 8, the dated change, starts at 596 seconds. Everything in between is the conversation itself.

Here is the shape in plain words. A board is not a search box. You are not asking it to write a thing for you. You are asking five different standards to look at the same work and tell you where it falls short. That only works if they can see the work.

In the video, Nick connected the transcripts from the best episodes of his podcast, the show's real performance data, and his most recent changes. So the board actually knew the show, not just what he says about it. That one move is the difference between advice and noise.

This guide is sponsored by Zapier, and Zapier is the tool Nick used to wire the board to his data. You can build your own board with Zapier at the link in our sources. We are saying that once, plainly, so you know where the tool came from.

If you only take one thing from this page, take step 3 and step 8. Real data in, dated change out. The middle is the fun part, but those two ends are what make the hour pay.

What are AI agents and what can they actually do here?

In this build, each of the five board members is an agent with a role, not a task, and that is the whole trick. The board is not asked to complete a job. It is asked to judge work, each persona bringing a different standard.

Most people use AI like a vending machine. Put in a request, get out a draft. That is a task. A role is different. A role says who is speaking, what they care about, and what they are looking at. Then you hand it your work and ask what it thinks.

The five roles in this board are AI personas built on the public ideas of MrBeast, Joe Rogan, Naval Ravikant, Seth Godin, and David Goggins. Our source does not spell out a single concern for each one. What it does say is that they pull in different directions.

That is the pattern to copy when the thing you want is an opinion rather than an output: define who is speaking, what they care about, and what they are looking at. Write those three lines for each member and you have a board.

What can it do? It can read your real numbers and name the gap between what you say you are doing and what the data shows. It can push back when you make a claim about yourself. It cannot decide for you. In the video, Nick disagreed with one answer on camera, and that is the right instinct.

How do I make AI give honest feedback instead of agreeing with me?

Two moves did it: telling the board it was allowed to disagree and call us out, and giving it real data so it could contradict us with evidence instead of vibes. That is step 2 and step 3 in the table.

The default behavior of an AI advisor is to agree with you politely, which is useless. Left alone, it will praise your plan, soften the hard part, and send you off feeling good. That is a bad advisor and a worse mirror.

So say it out loud in the setup. Tell the board it has permission to disagree, to push, and to tell you when you are dodging. In the video, Nick said he wasn't censoring himself. The Rogan persona came back with a line about how almost everyone thinks they're not censoring, and you can spot it in the seams.

That line only lands because the board had the transcripts. It could point at the seams. Without the data, it would have been a clever sentence with nothing behind it.

There is a third move, and it is quieter. Ask questions that have a real answer you do not want to hear. A question like "is this good" invites praise. A question like "what am I protecting" does not.

When you claim something about yourself, let the board check it. If it just nods, your setup is wrong, not your work.

Why use five AI personas instead of one assistant?

Five personas give you a disagreement; one gives you a chorus of one. In the video, Nick chose five people whose ideas shape how a lot of us think, and built a persona for each.

The members are AI personas modeled on the public ideas of MrBeast, Joe Rogan, Naval Ravikant, Seth Godin, and David Goggins. None of those people took part, and none of them saw any of this. When we quote a board member here, we are quoting the persona's output, not the person.

Pick voices you already argue with in your head. You know what each one would say about your work, roughly, and that is the point. The persona makes that argument sharp and puts it next to four other arguments that do not match.

A board of identical agents returns one opinion three times. The value here comes from the personas pulling in different directions, so the useful signal is where they conflict, not where they agree.

Watch for the conflict. In the video, the Rogan persona talked about two worlds: the polished creator-business world and the raw conversation world. That tension is the real finding. A single assistant would have smoothed it over into one tidy plan.

Five is a good number because it breaks ties and still fits in one conversation. You could do three. More than five and the answers start to blur together.

How do Zapier AI agents work in this setup?

Zapier does the connecting work behind the scenes, and in this build it pulled the transcripts, the performance data, and the recent changes into the board. That is step 3 in the table, at 36 seconds in the video.

Think of it as plumbing. Your transcripts live in one place. Your view counts and watch time live in another. Your notes about what you changed last month live somewhere else. The board cannot read any of that on its own.

Zapier moves those pieces to the board so the board can see them. As Nick says in the video, it's the unsexy part: pulling the data and the transcripts so you can have a real conversation with the board.

Once that line is open, the conversation changes. When the Nate Soares episode came up, Nick could point at real watch-time numbers instead of a feeling. The board was reacting to what happened, not to a summary.

This guide is sponsored by Zapier, and the link to build your own board sits in our sources. If you use a different tool, the rule still holds: the board must see the work, not your description of the work.

Start small. One folder of transcripts and one export of numbers is plenty for a first session. You can add more once you know which questions you keep coming back to.

What questions should I ask an AI board about my business?

Ask the ones with teeth, and in the video Nick asks a short list of open questions plus one forced tradeoff. The open ones are what is next, which side am I censoring for, what version of myself am I protecting, what decision am I avoiding because failing would expose me, what scares me most, and what do I secretly want this thing to give me.

Start wide. The first question was simply what the next step for the show is, and all five answered from their own angle. That gives you a map of where the five standards disagree.

Then push on one answer instead of moving on. Nick asked the Rogan persona what it meant by two worlds, and the follow-up is where the useful stuff came out. The first answer is a headline. The second answer is the story.

Then go private. The board earns its keep on the questions you'd never post publicly. Nobody's feelings are at stake, so you can ask the thing you would not say to a friend over coffee.

At 326 seconds, Nick asked what decision he is avoiding because failing would expose him. At 452 seconds, he asked what scares him most. Those two moments are in the key moments table.

Write your questions before the session. In the moment, you will drift toward the easy ones.

Growth or depth: which should a creator pick?

This board split three for depth, one for the third option, and one for growth, so the call was depth. That vote happens at 261 seconds and is step 5 in the table.

Open questions get you essays. So Nick asked the board to tell him what to sacrifice and gave it three fixed options. The exact three words sit in the step 5 row of the steps table. Fixed options turn a long discussion into a count you can act on.

The reasoning matters as much as the tally. In the video, the Naval persona argued that depth plus trust compounds and growth is a lagging indicator, so build the asset and let attention arrive as a byproduct. The Godin persona said to earn the smallest audience that cares more than they scroll.

That is this board on this show, not a universal answer. Your data is different, your personas may be different, and the vote could land the other way. The method travels; the result does not.

Counting the votes turned five opinions into one decision, so the board said go for depth. That sentence is the whole reason to force a tradeoff. Five smart takes with no tally leaves you exactly where you started.

Use fixed options. If you let the board invent its own fourth path, it will, and you will be reading essays again.

What is quiet irrelevance, and why does it sting more than failing?

Quiet irrelevance is slowly disappearing while still working hard, and it is the first moment in the video, at 0 seconds. That phrase came out of the board and it hit harder than any tactic in the conversation.

Being wrong in public is loud. People see it, you fix it, you move on. Quiet irrelevance has no moment. You keep shipping, the room keeps emptying, and nothing ever tells you to stop.

The Goggins persona framed the same thing as embarrassment now or regret later, and later is the scarier of the two. That is a useful frame for any family or business weighing a risky change.

At 207 seconds, the question is what version of myself am I protecting. At 326 seconds, it is the decision I am avoiding. Those two sit right next to the fear of quiet irrelevance, and they are the same problem from different sides.

The Godin persona said Nick wants permission to matter, not attention but resonance. That landed harder than any tactic in the conversation. It is the kind of line you only get when you let a machine say the blunt thing.

You do not have to agree with any of it. But if a sentence sticks in your head for a week, treat that as data.

How do I build multi-agent teams that work together?

Give each agent a distinct standard and let them disagree, because a board of identical agents returns one opinion three times. That is the core rule, and it applies well beyond a podcast.

Working together does not mean agreeing. A real board of five people would argue. Your AI board should too. If all five answers sound alike, you wrote one persona five times.

So write each role with a different thing it cares about, in your own words. Pick five standards that do not match. Then hand all five the same work and the same question.

Read the conflict, not the consensus. The value here comes from the personas pulling in different directions, so the useful signal is where they conflict, not where they agree. In the video, the clearest finding was a tension, not a plan.

When you need a decision, add a vote. Fixed options, one answer each, count the result. Three said depth, one said the third option, and one said growth, and that was the call.

Then do step 7. Disagree out loud when it is wrong. Nick told the Goggins persona on camera that he disagreed, and treating the output as a verdict defeats the point.

Is an AI board of advisors useful or just a gimmick?

In the video, Nick says it was useful, but not the way he expected, and that some of the advice was plain wrong. He said on camera that he disagrees with the Goggins persona at 381 seconds, which is step 7 in the table.

Here is the honest version. The advice itself was mixed. Some of it was generic. Some of it read like a motivational poster. If you go in hoping for a secret plan, you will be let down.

What worked was hearing an uncomfortable thing said plainly, from data about his own work, in a place where nobody's feelings were at stake. A couple of things hit him in the gut and he can't unhear them.

That is a fair test for any tool like this. Did one sentence change what you do next week? If yes, it did its job. If no, your setup probably skipped step 2 or step 3.

This isn't an oracle. Argue back, and keep the two or three things you can't unhear. The board gives you material. You still make the call.

The gimmick version is five chatbots praising you. The real version has your numbers in it and tells you something you did not want to hear.

What do I do after the session so the advice sticks?

Turn what hit you into a dated change, which is step 8 at 596 seconds. In the video, Nick committed to 30 days of changes: tighter structure, tighter intros, and cutting anything that doesn't earn its spot.

The breakthrough wasn't the advice, it was the system that turned all of it into action. A conversation with no date attached fades by Friday. A conversation that ends with a change and a deadline is a plan.

Pick two or three things, not ten. Nick picked three, and all three are about cutting, not adding. That is usually the right shape after a session like this, because the board tends to find padding before it finds gaps.

Say it where people can see it. If it works, you'll feel it. If it doesn't, you get to watch me be wrong in public, which beats quietly disappearing.

Then come back. Feed the board the new numbers after the 30 days and ask whether the change did anything. Because the board is wired to real data, it can tell you, and it will not be kind about it.

That loop is the whole product. Data in, hard questions, a vote, a dated change, new data. The personas are the fun part, but the loop is what moves the work.

Am I censoring myself, and can a board tell me?

It can, and in the video it did at step 2, 29 seconds in. Nick said he wasn't censoring himself. The Rogan persona came back with a line about how almost everyone thinks they're not censoring, and you can spot it in the seams.

That is a hard thing to hear and an easy thing to test. The seams are the places where your work goes vague. A topic you skirt. A story you tell up to a point and then cut off.

A board with your transcripts can find those spots because it has read them. A board with no data can only guess. This is step 3 doing its job inside step 2.

The same pattern showed up at 130 seconds, when the Rogan persona talked about two worlds. One is the polished creator-business world. The other is the raw conversation world. Censoring often means hiding one of those from the other.

For a family reading this, the lesson is simple. If you want honest feedback, give the reviewer the real work and real permission to be blunt. Then listen to the part you flinch at.

Method and sources: The data on this page comes from one video published on The Nick Standlea Show on March 9, 2026, and the page is current as of October 2026. The steps table holds all 8 steps as Nick lays them out, and the key moments table holds all 10 timestamped moments, both copied from the video in order. The board members are AI personas modeled on public ideas, so every quoted line is persona output, not a statement from MrBeast, Joe Rogan, Naval Ravikant, Seth Godin, or David Goggins. The video refers to real watch-time numbers for one episode, but no watch-time or view-count figures are reported in our source material, so none appear on this page. The guide is sponsored by Zapier, which is the tool used to connect the board to the show's data.