How to Automate a Board of Advisors for $29 a Month

Show Notes

Broke and scared has company. Everybody has been broke and scared. What nobody prepares you for is the 11 PM decision, three years in, revenue real, team growing, and not a single person left who understands the shape of the problem.

Funding stage: Bootstrapped. No disclosed outside round for CEOfriend, no listed investors, no cap table signal anywhere public. Revenue comes from a $29-per-user subscription, the platform was built with Clock Ltd as a digital partner rather than a venture-funded engineering team, and Steve is financing this from a career that has already returned over $200 million in raised capital elsewhere. This is a bootstrapped second act, not a pre-seed.

Your friends stopped relating around round two. Your family cannot parse the question. Your team needs you to already know. Steve Phillips solved that problem the hard way, six times.

He is the founder of Zappi, the consumer insights platform he launched in 2012 as ZappiStore, which raised $170 million in growth capital from Sumeru Equity Partners in December 2022, employs more than 300 people across 13 countries, serves more than 350 enterprise customers, and is approaching $100 million in ARR by his own account. He handed the CEO seat to Aaron Kechley in October 2024 and moved into Executive Chair and Chief Innovation Officer.

Then he did the thing almost no exited founder does. Instead of collecting non-executive director seats and giving advice people were free to ignore, he automated the role. CEOfriend.ai gives founders a working advisory team: a CFO agent modeled on the conservative finance people Steve has worked with for two decades, an entrepreneur agent that gets genuinely excited about your bad idea, an investor agent that keeps asking where the enterprise value is, plus marketing and sales and AI readiness advisors. Seven days free, then $29 per user per month on an introductory offer. The mission Steve states out loud: make every small business 20% better.

This episode goes deep on the mechanics of that build, the psychology of stepping down from your own company, why the pre-launch phase feels so good and is so dangerous, and a morning ritual involving bare feet, wet grass, and a cup of coffee in a Camden garden.

Named Frameworks

Founder First, CEO Second

Steve stood in front of the whole Zappi company once a year and said it out loud, alongside his CTO.

›The founder role is broad direction, conviction, and where the company is charging.
›The CEO role is operating the machine, and that is a separate skill that changes as the company changes.
›He publicly committed to re-evaluating every year whether he was still the right CEO for the next year.
›His honest self-assessment: strong CEO up to roughly $20 million, weaker past that, because org design and systematization were not where his head naturally goes.
›The outcome was not an exit. He stayed as Executive Chair and Chief Innovation Officer, with nobody reporting to him.

The Pirate Test

Steve's model for early customers and early feedback.

›Your first users are pirates by nature, people who ignore the standard way of doing things and try the unfinished thing on purpose.
›Be aggressively honest about where you are in product development, because pirates recalibrate their expectations when you do.
›They will hand you flavors you did not think of, and they will tell you when it sucks.
›Falling in love with this stage is the antidote to falling in love with the solution.
›Pirates have to be cultivated deliberately. Steve keeps a standing group of seven or eight entrepreneurs on a monthly Zoom for exactly this purpose.

The Conflicting Counsel Model

The design principle behind the CEOfriend personas. Steve built them to disagree with each other on purpose.

›The CFO agent's job is to tell you that you cannot afford it, working off your P&L and your three month cash position.
›The entrepreneur agent's job is to love the idea and pile on four more.
›The investor agent's job is to ask how this creates enterprise value, what the ROI profile looks like, and whether you will need more money.
›The value is in the spread between them, not in any single answer.
›Each persona was modeled on real people Steve has worked with over 15 to 20 years, which is the part a generic prompt cannot replicate.

Work on the Business, Not in the Business

The corrective to a founder's default state of urgency.

›The default founder state is deep in urgent things: the customer call, the moaning employee, the hire, the bill, the bank.
›All of it is genuinely important, which is exactly why it eats strategy.
›The corrective is a rolling 30, 60, 90 day plan that you can edit, argue with, and tick off.
›The output should be a top three, not a top 37.

The 20% Better Mission

CEOfriend's stated reason for existing.

›Big companies can afford hundreds of thousands of dollars in real advisors, and should go spend it.
›Small businesses are the backbone, and they get nothing comparable.
›The bet: point frontier models specifically at small business decisions rather than building another general assistant.
›The price ceiling is deliberate, because the mission fails if a first-time founder cannot afford it.

The Barefoot Boot-Up

Steve's mental health stack, offered without any claim that the science holds up.

›Five to ten minutes standing barefoot in the garden every morning with a coffee.
›Watch what is blooming, what is decaying, which birds arrived, which flowers the bees went for.
›Walking meetings for the last ten years, with staff, clients, and business partners, on mapped routes through Regent's Park.
›Ten miles a day cycling in London bus lanes, 10,000 steps minimum, 25,000 on a cricket day.
›The theory about electrons is his friend's, and Steve is openly unconvinced. He keeps doing it because it works on him.

Key Terms

ARR: Annual recurring revenue. The annualized value of subscription contracts currently in force.

Non-executive director (NED): A board member with no operational role, common in UK governance. Advises, votes, and, as Steve puts it, can be ignored.

PE backed: Private equity backed. Ownership sits with a fund rather than founders or venture investors, which typically changes the reporting cadence and the exit clock.

Growth capital: Later-stage funding for companies with proven revenue, usually less dilutive per dollar than early venture and often structured with a liquidity component for existing holders.

P&L: Profit and loss statement. The document the CFO agent reads before it tells you no.

SWOT: Strengths, weaknesses, opportunities, threats. The frame CEOfriend's free competitor report is built around.

Competitive intelligence report: Structured analysis of your market position against named competitors, including adjacent entrants you are not currently watching.

Agentic swarm: Multiple AI agents with distinct roles and instructions working on the same problem, rather than one general assistant.

Persona agent: An agent shaped by a role's incentives and communication style, so its output carries the bias of a real job function.

Model routing: Sending different tasks to different underlying models based on which performs best on that task type. CEOfriend routes across Gemini, ChatGPT, Claude, and Grok.

Human in the loop: A person supplying judgment, taste, or correction inside an automated system. Steve's version of this is the founder experience baked into each persona.

Zero to one: The pre-product-market-fit stage. Steve's favorite part, and the reason he keeps starting companies.

Pirates: Steve's term for early adopters who will use an unfinished product and tell you the truth about it.

30, 60, 90 day plan: A rolling strategic plan on three horizons, editable and checkable, designed to survive contact with the daily chaos.

B Corp: A certification for companies meeting verified social and environmental performance standards. Zappi holds one.

Grounding: Standing barefoot on soil or grass. The claimed mechanism involves electron transfer from the earth. The claim is not established science, and Steve says so himself.

Q&A

What is CEOfriend.ai and how much does it cost?

CEOfriend.ai is an AI advisory platform for founders and small business leaders, built by Zappi founder Steve Phillips. It gives you a set of specialist AI advisors including a CFO, an entrepreneur, an investor, a marketing and sales advisor, and an AI readiness advisor, plus competitor analysis, board pack optimization, and market trend reports. It runs seven days free, then $29 per user per month on an introductory offer, down from a $49 list price. The competitor analysis is free with no signup required at ceofriend.ai.

Why not just build an AI advisory board yourself in ChatGPT?

Steve's answer, from someone who tried it: you can, and it will take a lot of painstaking back and forth to get it working well, and then you are locked to one model. CEOfriend routes across multiple models based on which performs best per task. The deeper argument is time allocation. Most founders want to work on their own business, not rebuild somebody else's.

How do you know when to step down as CEO of your own company?

Steve's method is a public annual re-evaluation. He told the whole company every year that he had two roles, founder and CEO, that founder came first, and that he would assess each year whether he was still the right CEO. His conclusion was that he was strong up to roughly $20 million in revenue and weaker past that, because org design and systematization were not his natural strengths. He handed the CEO role to Aaron Kechley in 2024 and stayed on as Executive Chair and Chief Innovation Officer.

How do you use AI to make better strategic decisions as a small business owner?

Point it at the strategic work you keep deferring rather than the tactical work already on fire. Steve's design gives you a rolling 30, 60, 90 day plan, proactive market trend updates you did not request, and a persona set that argues with each other so you get the spread rather than one agreeable answer. The framing he uses: work on the business, not in the business.

What makes a good early customer for a startup?

Steve calls them pirates. They ignore the standard way of doing things, they will use a buggy product on purpose, they expect imperfection when you are honest about the stage you are at, and they give real feedback including the version you do not want. Cultivate them deliberately rather than waiting for them to appear.

How do founders deal with isolation after they get traction?

Steve's structural answer is a standing group of seven or eight entrepreneurs who have known him a long time, meeting monthly on Zoom and quarterly in person, who will say "this sucks" without softening it. His product answer is CEOfriend, which he built specifically because that kind of group is rare and most founders never assemble one.

Is the UK a good place to build an AI startup?

Steve's read on London: DeepMind is there, a large share of Google's AI engineering talent sits in the city, the science base is strong, and the entrepreneurial scene is flourishing, though nowhere near the scale of the United States. What the UK still gets wrong is treating failure as a verdict instead of a data point, though he thinks the American reframe is seeping in.

I have spent months polishing and still have not launched. What do I do?

Recognize what is actually happening. The pre-launch phase feels perfect because nothing has gone wrong yet, and nothing has gone wrong because nobody has seen it. Steve's reframe is to fall in love with the feedback stage instead of the solution stage. Tell people plainly that you are early and want feedback, and they will show up as collaborators rather than judges. His cautionary tale in this episode: an engineer who spent a quarter of a million dollars building a $100 toothpaste tube squeezer without ever asking whether anyone would pay for one.

Should I build my own AI tooling or buy someone else's?

Ask the time versus value question honestly. Steve's answer for his own category: yes, you could rebuild CEOfriend in Grok, and it would take a lot of painstaking iteration, and then you would be locked to Grok while the frontier keeps moving. If the tool is your product, build it. If the tool supports your product, the hours you spend rebuilding it are hours not spent on the thing customers pay you for.