AI Mentor vs an 'Act as Rob Walling' Prompt
· MentorClone Team
You've probably tried it. Open ChatGPT, type "Act as Rob Walling and tell me whether I should raise my prices," and get back a warm, confident paragraph that sounds a lot like a podcast host who has helped hundreds of SaaS companies.
It sounds a lot like Rob. You have no way to know if any of it is his.
What the prompt actually does
A general chatbot has read a huge amount of text. Ask it to act as someone and it predicts what they'd plausibly say, based on everything it has seen about them and everyone who sounds like them. You get a blend.
For a real decision, the blend has problems. You can't tell which sentences came from Rob, from a blog post quoting Rob, from a different founder with a similar style, or from nowhere. There's no episode to go back to. And it stays in character, so if you ask about something Rob never covered, you get an answer anyway, in his voice.
ChatGPT is good at drafting and explaining. The problem is narrow. When you want one specific person's advice, a persona prompt can't show you that the advice is theirs.
The same question, asked of a mentor
MentorClone builds each mentor from one person's own work. Right now that's Rob Walling, from his Startups for the Rest of Us episodes, and Arvid Kahl, from The Bootstrapped Founder articles and his videos.
The mentor answers only from that material. You tell it your situation and it tells you what to do. Each point links to where they said it: the exact moment for video, the page for an article or episode. If the person never covered your question, it says so. It also remembers your situation for next time.
Take a question people really ask: "I run a small B2B SaaS. Nobody has complained about my price in a year. Should I raise it?"
Here's what sits behind the answer on Rob's pricing topic page, word for word:
"if you're not getting any complaints about price, then your price is too low"
Startups for the Rest of Us, Episode 702 (February 2024)
"a lot of folks are undercharging when they start with TinySeed. And of all the times that TinySeed companies have decided to raise their pricing, I believe I've only seen one maybe two times where that was a mistake"
Episode 785 (July 2025)
And from the same episode, the caveat a generic answer tends to skip:
"they went, not even upmarket, but they just raised their pricing 20, 30, 40% and it was a mistake and it kind of wrecked their funnel and they realized, oh, we outpaced the market"
Anyone can tell you to raise your prices. Rob also says how he's seen it go wrong, and you can open the episode where he said it.
Arvid's pricing topic page adds what he did at his own company:
"Here is what we did at FeedbackPanda when we increased our prices by 50%: we allowed our legacy customers to keep their old prices for a year and then adjusted them to the new price."
The Bootstrapped Founder, "Pricing for Calm SaaS Businesses" (October 2022)
A persona prompt might produce something close. The mentor shows you the article, so you can read the whole thing before you email your customers. Both founders' evidence is laid out in Should You Raise Your SaaS Prices?.
Dates matter
People change their minds. A persona prompt flattens ten years of someone's thinking into one voice. The mentor keeps the dates, and topic pages show how a view moved. When Arvid writes in 2020 that "I won't make the mistake of grandfathering users indefinitely again this time around," you can see what he learned from, and check the date yourself.
The most useful answer is "no source"
Ask Rob's mentor a medical question, or about a legal detail in your country, and it will tell you it has nothing from Rob on that. The persona prompt answers anyway.
That plain "not covered" sends you off to find the right person before you act on a guess in a familiar voice.
Which one to use
Use a general chatbot to brainstorm, draft, or get a feel for common practice, when you don't need to know who said what. Use a mentor when you want one person's view on your decision and you want to check it before you act.
Plenty of people do both. Draft with the chatbot, then check the decision against someone whose track record you trust. Your AI agent can ask the mentor too. See Give Your AI Agent a Mentor.
Doing it yourself
You can get part of the way. Paste a few articles or transcripts into a chatbot, tell it to answer only from them, and ask for quotes. For a handful of sources, that works.
It falls apart at the size of a real body of work. Rob has recorded hundreds of podcast episodes. Arvid has written hundreds of articles. Keeping every point tied to its exact source, dated and checked against the passage it came from, is the work we do before you ask anything. For the difference between a notebook of your own files and a ready-made mentor, see MentorClone vs NotebookLM.
Frequently Asked Questions
Is an AI mentor the same as a ChatGPT persona prompt?
No. A persona prompt asks a general model to imitate someone. An AI mentor answers only from that person's own published work, links each point to where they said it, and tells you when they never covered your question.
Does MentorClone speak for Rob Walling or Arvid Kahl?
No. Each mentor is built from the person's public work. They didn't write or endorse it. Every answer links back to their own words so you can judge for yourself.
What happens if I ask about something they never covered?
The mentor tells you it has nothing from that person on your question. It doesn't fill the gap with general knowledge.
How much does it cost?
Plans start at $19 a month, and you can try 3 questions on each mentor page first. See pricing.
Try it on your own decision
Ask Rob Walling or Arvid Kahl the question you last gave ChatGPT. Put the two answers side by side, then open the sources.