I’m Glenn Gow, The Scaling Executive Coach and host of The Scaling Executive Podcast. In this episode, I sat down with Howard Chang, CEO of theturnlab and Just Meeting Rooms, to talk about how he uses AI inside two companies at once. Howard’s rule is simple: when AI gives you a complex answer, ask it “are you sure?” before you act on it. That single follow-up question changes the answer roughly half the time, which tells you how often a first AI response is wrong in ways you would not otherwise catch.
This episode is for CEOs who are already using AI in daily decisions and need a concrete check to catch the errors before they become costly ones.
Key Takeaways
- Howard Chang, CEO of theturnlab, built a standing rule for his team: challenge every complex AI answer with a second prompt, “are you sure?”, because the first answer frequently does not survive that check.
- Chang treats AI as a “useful idiot”: fast and genuinely helpful, but prone to getting simple things wrong without a verification step.
- theturnlab is shifting some creative production work directly to clients using AI platforms, moving the agency from doing the hands-on work to guiding the strategy behind it.
- Chang accepts that this shift may shrink theturnlab’s short-term revenue because he prioritizes building the right business over building the biggest one.
How Should a CEO Verify an AI Answer Before Trusting It?
Howard Chang built his verification habit around a single follow-up question. When his team asks Claude or ChatGPT a complex question, he tells them not to stop at the first answer. He requires a second prompt: “are you sure?”
“Are you sure? And interestingly enough, 50% of the time changes the answer.” — Howard Chang
That statistic is the reason the habit exists. If an AI tool changes its own answer half the time when challenged, then treating the first answer as final is a coin flip on accuracy. Chang’s fix costs nothing and takes seconds: build the second prompt into the workflow as a standing rule, not an optional step.
This is also why Chang frames AI the way he does. He does not treat it as an oracle and he does not dismiss it as a toy.
“I see AI as kind of like the useful idiot.” — Howard Chang
The label sets the right expectation for a CEO managing a team that uses AI daily. Useful means it earns a place in the workflow. Idiot means it still needs a human check before its output reaches a client or a decision.
How Can CEOs Use AI to Shift Their Teams from Doing the Work to Guiding It?
Howard Chang, CEO of theturnlab, is using AI to change what his agency sells. theturnlab historically produced creative work directly for clients. Now, on at least one account, the agency is moving some of that production to the client itself, using an AI platform that lets the client’s own team execute work that used to require theturnlab’s staff.
Chang does not see this as losing the account. He sees it as repositioning what theturnlab is paid for.
“It’s allowing us to be more the heads than the hands for a client.” — Howard Chang
Under this model, theturnlab keeps the strategic judgment work: diagnosing the problem, setting direction, deciding what to build. The client’s team, supported by AI tools, handles the repeatable production work that used to require a full agency staff. For a CEO running a services business, this is a specific, actionable pattern: identify which parts of your delivery are judgment and which parts are repeatable execution, then use AI to hand the repeatable parts to the client while your team holds onto the judgment.
Should a CEO Accept Lower Revenue to Build a Stronger AI-Enabled Business?
Handing production work to clients has an obvious downside: it can shrink the revenue theturnlab used to earn from doing that work itself. Chang has made peace with that trade explicitly.
“If this is about building the right business, not building a big business.” — Howard Chang
Chang is naming a real trade-off, not avoiding one. Short-term, this AI-driven shift can reduce theturnlab’s top-line revenue on affected accounts. Chang’s position is that a business built around durable client trust and high-judgment work is worth more over time than one built around billable hours that AI can now compress. For a CEO evaluating whether to let AI eliminate part of their own service delivery, the decision hinges on which asset they are actually protecting: revenue in the current quarter, or the client relationship and margin structure over the next several years.
How Is theturnlab Restructuring Around AI, Revenue, and Client Judgment?
| Principle | What it means in practice | Named evidence from this interview |
| Verify before you act | Require a second prompt (“are you sure?”) on any complex AI-generated answer before using it in a decision | Chang reports the answer changes roughly 50% of the time when his team challenges the first response |
| Reposition the team around judgment | Move repeatable production work to AI and the client, and keep your team focused on strategy and diagnosis | theturnlab grew from 12 employees at founding to 50 today under this consultancy-first, “heads not hands” model, and Chang is now extending that same reasoning to the AI production handoff on at least one client account |
| Protect the business, not just the revenue line | Accept short-term revenue compression from AI adoption if it strengthens long-term client value and margin | Chang has made this trade before: in 2016 he closed Top Drawer Creative, a profitable 20-plus-year, 60-person agency he had built, to found theturnlab ahead of the shift he saw coming — the same long-view calculus he is applying to this account now |
Quotes from This Episode
- “AI hallucinates, AI often gets things wrong.” — Howard Chang, CEO, theturnlab
- “You can gain some artificial data as a starting point quickly, but you absolutely need to verify because it does hallucinate.” — Howard Chang, CEO, theturnlab
- “In the short term, it might actually shrink some of our top line revenue. I don’t care.” — Howard Chang, CEO, theturnlab
Frequently Asked Questions
How should a CEO evaluate an AI-generated answer before acting on it?
Howard Chang, CEO of theturnlab, requires his team to follow up any complex AI answer with a direct challenge: “are you sure?” He reports that this second prompt frequently changes the AI’s answer, which means the first answer alone is not a reliable basis for a business decision without that additional check.
Can AI actually replace expert judgment in a service business?
Not according to Chang. He treats AI as a “useful idiot,” meaning it speeds up production and drafting but still makes basic mistakes without human verification. At theturnlab, AI is used to shift repeatable execution work to clients, while the agency’s own team keeps the strategic judgment work that AI cannot reliably replace.
Should a CEO worry about revenue dropping when they adopt AI-driven service models?
Chang expects some near-term revenue compression as theturnlab shifts creative production to clients using AI tools, and he accepts that trade-off. His reasoning is that building a durable, high-judgment business is worth more over time than protecting a revenue line tied to hours of production work AI can now handle.
CEOs Work with Glenn Gow to Turn Fast AI Answers Into Decisions They Can Trust
Glenn Gow is The Scaling Executive Coach, coaching ambitious executives into the CEO seat and CEOs into successful exits. With 25 years as a CEO and 5 years in venture capital, Glenn helps leaders build the habits that keep AI-driven decisions accurate as they scale. If this conversation was useful, you can apply for executive coaching with Glenn Gow or apply to be a guest on The Scaling Executive Podcast.
