For years, founders have been told the same thing: build a defensible moat. Proprietary tech. Network effects. Lock-in.
But in AI, today’s moat can disappear with the next platform update.
That tension is at the center of my conversation with Ang Li, co-founder and CEO of Simular, a company building AI agents that operate using the same tools humans do: mouse, keyboard, and screen.
This isn’t a conversation about how agents will change the world. It’s about how founders make decisions and build companies when certainty isn’t available.
The Bigger Idea
Ang doesn’t argue that moats don’t matter. He argues that in fast-moving technical markets, static moats don’t exist. What replaces them is a culture of shipping, tight feedback loops, and competing against your own past performance, not just your competitors.
As Ang puts it:
“The hidden part is the real moat, meaning everyone in the company has a mindset to innovate consistently and learn from the feedback consistently — continually learning, continually innovating.”
"There's never a moat on anything."
A Few Tactical Takeaways
Rather than preach theory, Ang was unusually concrete about how he operates day to day:
Speed is about feedback, not velocity.
Simular ships prototypes in one- to two-week cycles — not to “move fast,” but to force real customer feedback as early as possible.Don’t default to hiring if automation can remove the work entirely.
Ang’s rule of thumb: if he’s spending most of his time on a task, that’s a signal to either automate it, or hire someone who’s better than him at it.There is no permanent technical moat.
Data, models, benchmarks, and even users are temporary advantages. The only durable edge is a team that adapts faster than the environment changes.
Note: This interview was recorded before Simular closed its $21.5M Series A in December 2025.
RUNTIME 56:44
EPISODE BREAKDOWN
(1:52) What is Simular, and how does it work?
(6:11) How Ang and co-founder Jiachen Yang connected
(9:00) How much time passed between Day Zero and serving their first customer?
(13:54) The moment Ang realized " this is gonna be like something huge."
(17:21) How he approaches founder-led sales and what he looks for in a GTM hire
(26:34) Maintaining cohesion when you're leading a distributed team
(32:23) Should you hire a new employee, or build a new agent?
(34:50) Why Ang made talking AI gorillas part of Simular's GTM strategy
(38:20)"If everyone becomes too cautious there, that actually prevents the innovation part."
(43:55) "There's never a moat on anything."
(51:16) The final question
LINKS
Meet the AI Agent with Multiple Personalities, Wired, 4/16/2025
Stay in the Conversation
1. Get the Fund/Build/Scale newsletter on Beehiiv
I moved everything off Substack so I can centralize here.
2. Subscribe wherever you get your podcasts
3. Want to support this work?
Become a paid supporter. It directly buys me time to make more deeply-reported, founder-first conversations like this one.
Thanks for reading,
— Walter.
