Our customers are often amazed and ask us about how our team ships this fast. The boring answer is that we almost never debate about things that don't need it. The way we pull it off is a sorting trick: every product decision goes into one of three buckets.
Hats
You can take a hat off and put on a different one whenever you want. These decisions are reversible in minutes, so we don't debate them. We ship, then let customers tell us if we got it wrong. We named the human-review step in our Agent Builder "Approvals." Perfect word? No idea. We shipped it. If users get confused, we rename it. Nobody needed a meeting. Mukul shipped a hover card on the customer list view without asking anyone, because it's a hat. If it was wrong we'd pull it. It wasn't.
Haircuts
A bad haircut grows out. It's annoying for a few weeks, but you're not stuck with it forever. These are medium-term bets. We discuss them briefly, pick a direction, and move. We can change course later, it just costs us something. Our integration strategy is a haircut: Ampersand for structured data like Salesforce, Stripe, and HubSpot, Composio for action-based stuff like Gmail, Slack, and Linear. Could we migrate off either one? Yes. Would it hurt? Also yes. So we thought about it for an afternoon, not a quarter. Same with Inngest as our execution layer for agents. Real commitment, replaceable commitment.
Tattoos
You live with a tattoo. Removing one is painful and expensive. These get genuine deliberation, because everything else gets built on top of them. We build agents as LLMs with tools, not as rigid DAGs. That's a tattoo. It shapes every feature we've shipped since. We argued hard about it, and we should have. Our unified customer data model is the same. Get the core primitive wrong and you're re-platforming a year later.
The payoff
Speed: most decisions are hats. Once your team internalizes that, they stop holding meetings about hats.
Autonomy: engineers ship hats without asking, flag haircuts, and escalate tattoos. Everyone knows the rule, so nobody waits on me to move.
Focus: we save our debates for the tattoos, which is exactly where debates belong.
Most teams move slow because they tattoo everything. We move fast because we know most decisions wash out in the next shower.