Eric Lubow

5/ The same culture that accelerates experimentation is what makes the org governable again. Compliance teams who get to operate downstream of a sharing culture have a much easier job than ones who have to discover everything when it's too late. eric.lubow.org/2026/if-you-wa…

4/ The enables AI-native operations. Not "we use Claude for coding," that's AI-enabled. AI-native means the organization can learn from its own activity at the speed of AI. Each round of building gets cheaper because the previous rounds are accessible and searchable at build time

1/ Celebration sounds soft until you realize it's functioning as governance infrastructure. When anyone can ship working software, governance stops being a process problem and becomes a visibility problem. Visibility doesn't come from tools, it has to be baked into culture.

@GergelyOrosz It's amazing to me that people think this is a problem. This is literally user research. Not everyone has the same use cases & asking helps to uncover new ideas. It's also the basis of product mgmt. Asking what one can do better is peak vulnerability & needed for self-improvement

It’s incredible rare (and hard!) to find people like this Some of the success of Claude Code, Codex and OpenCode is surely that they have more than one such dev, who talks directly with users, including in public. Then does fixes based on what they learn!!

There are few things as lame as dunking on an engineer asking for feedback, esp in public. Honestly, a dev that has asked for feedback *once* in their lives, *in public* is already top 1% or above. Doing it regularly is somewhere too 0.0X% It adds up+makes big differences

Celebration sounds soft until you realize it's functioning as governance infrastructure. New piece on the cultural mechanisms that make AI-enabled orgs governable- the channels, meetings, & agents that turn visibility into the precondition for compliance: eric.lubow.org/2026/if-you-wa…

7/ Distributed building creates more coordination surfaces, not fewer. The PM role didn't disappear, it moved from gatekeeping engineering time to applying judgment where it matters. Full piece, including the accountability shift and what to stop doing: eric.lubow.org/2026/what-is-p…

6/ Models can write a PRD, build a prototype, draft a changelog, summarize a user interview. What they can't do is get 3 people in a room to agree. Cross-functional coordination is the work that can't be automated and it's now more important than ever.

1/ If everyone can build with AI, what's left for product management to do? That's the question I've been sitting with. The answer isn't "nothing" -- the frontier AI companies are hiring PMs aggressively, not shedding them. But the job has shifted in some specific ways.

The frontier AI companies are hiring PMs aggressively. Interesting signal at a moment when everyone's asking whether PM survives AI. New piece on what product management is actually for when everyone builds: eric.lubow.org/2026/what-is-p…

5/ The piece covers who's responsible for recognizing the transition, a plain-language checklist any builder can use, and what to do about it — from a lightweight coaching conversation to a full engineering hand-off. eric.lubow.org/2026/infrastru…

4/ The new AI-Engineering first principle: when other things depend on what you've built, it needs engineering judgment applied — regardless of who built it. Not to slow things down, but to make sure the foundation can carry the weight.

1/ The hardest infrastructure problems right now aren't about what to build. They're about recognizing what already became infrastructure while nobody was looking.