Why This CTO Subscribes to Three AI Coding Tools at Once

Three subscriptions. Not because he can’t pick a favorite.

Daryl Teo is a fractional CTO. He works across multiple clients, multiple tech stacks, multiple stages of product maturity. Every one of those clients depends on his judgment about which AI tools to adopt, which to skip, and which ones are pure marketing.

So he subscribes to all of them.

“I subscribe to Claude Code, Codex, and Cursor simultaneously. I need to know how they actually work, not just what the marketing says.”

— Episode 49, The AI Product Leader

Three coding tools. Three monthly bills. Not because Daryl is indecisive. Because his clients are making bets worth millions of dollars based on his recommendations, and he refuses to make those recommendations from a distance.

The cost of real judgment

Most technology leaders evaluate tools by reading the docs, watching the launch video, maybe trying the free tier for twenty minutes. Daryl treats it like fieldwork. He lives inside each tool long enough to feel where it breaks, where it shines, where the marketing diverges from the reality.

That distinction matters more than it sounds. Claude Code handles one type of workflow beautifully. Cursor excels at something else. Codex has its own strengths and blind spots. You can read comparison articles all day and still not know which one will fall apart on your specific codebase, with your specific constraints. The only way to know is to use them all, seriously, on real work.

Daryl’s philosophy on this is blunt:

“Lean in and commit. That’s my whole philosophy with AI. You can’t half-do it and expect to understand it.”

The “half-do it” part is what I see constantly. Leaders who tried one tool for a week, hit a friction point, and concluded that AI coding tools aren’t ready yet. They formed an opinion from a sample size of one bad afternoon. Daryl formed his opinion from months of daily use across three platforms simultaneously.

Compromised judgment

The sharpest thing Daryl said in our conversation wasn’t about tools or subscriptions. It was about what happens when you don’t do this:

“If you’re leading technology and you’re not using Claude Code or something like it personally, your judgment is compromised.”

Not “slightly behind.” Not “missing some context.” Compromised. That’s the word he chose, and he meant it. Your ability to evaluate your team’s work, to assess vendor claims, to make architecture decisions — all of it degrades when your understanding of the tools is secondhand.

This is the AI player-coach argument distilled to its most uncomfortable version. You can be brilliant. You can have twenty years of experience. If you haven’t personally wrestled with these tools, you’re operating on borrowed understanding. And borrowed understanding has an expiration date that gets shorter every month.

What three subscriptions actually buy

Daryl isn’t collecting tools like trading cards. Each subscription buys him something specific: the ability to walk into a client engagement and say “I’ve built with this, I know where it works, I know where it doesn’t, and here’s what I’d recommend for your situation.” No hedging. No “I’ve heard good things.” Direct experience converted into trustworthy guidance.

For a fractional CTO, that trust is the entire product. His clients aren’t hiring him for his resume. They’re hiring him for his judgment. And his judgment is only as good as his last hands-on session.

The three subscriptions cost maybe $100 a month combined. The cost of giving a client bad advice because you relied on someone else’s benchmark? Incalculably higher.

The pattern I keep seeing

Daryl’s version is extreme, but the principle shows up in every strong technology leader I talk to on the podcast. The ones making the best calls are the ones with the freshest firsthand experience. They don’t outsource their understanding to blog posts, analyst reports, or their most technical direct report.

They commit. They lean in. They accept that staying current with AI tools is now a recurring cost of doing the job well, not a one-time investment.

The leaders who treat tool fluency as someone else’s job are the ones whose judgment is quietly eroding underneath them. They just don’t know it yet.

Polly Allen is the founder of AI Career Boost and host of The AI Product Leader podcast. She spent years leading AI at Amazon Alexa before building the AI Career Boost Blueprint, an 8-week program for Director+ product leaders becoming indispensable AI player-coaches. Subscribe to The AI Player-Coach newsletter →