Janis Rozenblats

Product Leader | Scaling product orgs 0→1→N | Available for advisory | AI-first product development

[Janis Rozenblats @Janis_R]

@svpino Companies limiting tokens usually have a different problem - they're not sure AI is core to their process yet. If it is, $20k per engineer is cheap. If it's not, no token amount matters. What's your actual operating model choice?

[Janis Rozenblats @Janis_R]

@lennysan @OpenAI I've watched great engineers ship beautiful code toward wrong problems. Lenny's point lands - discipline is non-negotiable. What's changed: the person holding that discipline is now a builder who architects and ships code, not a strategy-only operator.

[Janis Rozenblats @Janis_R]

@svpino The companies winning aren't fighting sentiment - they're redesigning ops. Orgs that cut headcount without rebuilding lose talent. Orgs rebuilding workflows and distributing ownership keep people and ship faster. Operating model wins.

[Janis Rozenblats @Janis_R]

@svpino The ban doesn't solve it. You need operating model changes ... acceptance criteria, testing gates, quality ownership for AI output. Build the system first. Without it, developers cut corners human or AI.

[Janis Rozenblats @Janis_R]

@svpino Call center orgs watching this: the margin is real, but only if you restructure what you DO. Cut staff without redesigning operations and you lose the margin to attrition and retraining in 18 months. What's your actual operating model shift?

[Janis Rozenblats @Janis_R]

@svpino Speed is table-stakes now. Real blocker: content teams can't produce at the rate these models generate. Real-time tools need real-time org operations. How fast can your team iterate today?

[Janis Rozenblats @Janis_R]

Save this before your next PM hire. The best time to make these structural calls is before the team has already outgrown the workaround.

[Janis Rozenblats @Janis_R]

You're about to make your third PM hire. Are you building coordination capacity, or coordination debt that will stall you at Series B?

[Janis Rozenblats @Janis_R]

The decisions that stall you at Series B aren't hiring decisions. They're structural ones disguised as temporary workarounds. Every "we'll fix this when we're bigger" is a debt payment made at 3x the original cost.

[Janis Rozenblats @Janis_R]

@svpino Google's showing the architecture. But stateful agents are team ownership problems, not code problems. You can't distribute state without restructuring how your org validates and hands off work. Where do you hit friction?

[Janis Rozenblats @Janis_R]

@svpino Define the decision before the lock-in: Who chooses providers? What cost/output changes trigger a switch? Most teams don't ask until escape velocity's too high. Get the framework locked, then build the gateway.

[Janis Rozenblats @Janis_R]

Most CPOs think their job is getting devs onto AI tools. Faros AI measured 25-39% individual productivity gains with zero improvement in delivery velocity. Your devs got faster. Your org didn't. Converting saved hours into shipped outcomes: that's the job now.

[Janis Rozenblats @Janis_R]

@svpino @withneo Purpose-built infrastructure compounds a structural cost advantage. Most teams avoid specialization because it feels like a commitment. MCP proves the gap is real: you're paying 60% for optionality you never use.