Building the OS for knowledge work.
Pulse started between an operator and an engineer who’d watched startups scale from 50 to 100 to 300 people and watched the same thing happen every time: decision velocity collapses, the founder becomes a bottleneck, and the company’s pulse goes quiet. We sought to build the thing we wished we’d had, so that scale, handovers, and Friday-evening crises never break the operation again.
Origin
The company started in 2026 in a co-working space in Mumbai, in the gap between Claude Sonnet 4 and the first wave of AI agents that could actually use your tools. We’d been at companies where someone had bolted ChatGPT onto a sidebar and called it AI strategy. The pattern was always the same: impressive at the demo, ignored within two weeks.
The diagnosis was always the same too. The model was fine. The problem was that the assistant had no memory of yesterday, no understanding of the company’s vocabulary, no read on what mattered to this person in this role on this Tuesday. It was a brilliant intern with severe amnesia, asking you to brief it from scratch every time you opened a new tab.
So we wrote a manifesto. An assistant should know the company. It should remember. It should read the room. It should be auditable, governable, and ours. We called it Pulse because the thing we wanted was a sense of the heartbeat of work, what’s actually happening, who needs what, what’s about to break.
We are still small, still founder-led, still revising the manifesto. Pulse is in early access with a growing group of design partners; the public demo queue is how the next cohort gets in.
Read the full manifesto →How we operate
We are governance-first.
Every product decision starts with permissions and control. If a feature can’t be something the customer can scope, audit, and turn off, we don’t ship it.
We do not own data we cannot defend.
Customer data is held in escrow against a deletion certificate. The architecture assumes we will be subpoenaed and audited, and is designed so that the answers are simple.
We write before we build.
Every meaningful feature starts as a memo. If we can’t explain it in two pages, we haven’t understood it. Our memos are public to the team and most are public externally.
Long-running over chat-shaped.
Our customers’ work is patient and structured. We optimise for hour-long agent runs that ship reviewable work, not for snappy one-shot answers in a chat box.
Boring on infra, opinionated on product.
We use unglamorous, well-understood infrastructure (Postgres, Vercel, Supabase) and spend our novelty budget on the model layer and the permissions layer, where it earns its keep.
The founding team
Two co-founders who’ve been together since the first whiteboard. We hire slowly; the rest of the team is on the careers page.
Apoorv Jain
Operator background; product and growth at consumer + B2B SaaS companies. Writes most of our manifestos and reads every contact-form submission on Friday.
The numbers, plainly
Want to use Pulse? Want to build it with us?
Two doors. If you’re a team, talk to the founder. If you’re an engineer or designer who agrees with the manifesto, see careers.