Sam Seely

The workflow engine that nobody wanted to build

There is a stretch of every product engineering roadmap where a ticket appears titled something like "notification refactor" and everyone in the standup develops a sudden interest in their laptop. Somebody has to write it. Nobody wants to. It is unglamorous, cross-cutting, and the second it breaks, someone in customer support is asking why 40,000 users got the same welcome email at 3 a.m.

Sam Seely spent enough time watching this happen - first at Percolate, then at Frame.io, where he met his eventual co-founder Chris Bell - to conclude that the ticket was not the problem. The category was. In 2021 he and Bell started Knock, which is built to be the thing engineers reach for instead of writing the ticket.

The pitch is deliberately unromantic. Knock is an API. There is a dashboard, but he will tell you the dashboard is second. There are SDKs in every major language. There is a workflow engine underneath that does the interesting parts - batching, throttling, cross-channel routing, quiet hours, user preferences, the version control on the templates, the audit logs the enterprise buyer asks about on call three. In February 2024 Craft Ventures led a $12 million Series A into all of that, adding to a $6 million seed from Afore Capital that Seely had never publicly announced. Vercel is a customer. So are Amplitude, Hiive, and Betterworks. Twenty-eight people work there.

Seely's frame for it, given to TechCrunch, is that the best payments engineers now go work at Stripe and the best search engineers go work at Algolia, and there ought to be a place where the best notifications engineers go work. That place, he thinks, is Knock. It is a comparison founders make about their own companies all the time. Most of the time it is not true. In notifications, it is at least defensible.

Frame.io was the tell

Seely grew up going through the reasonable-sounding sequence of jobs that produces most founders in this category. Santa Clara University from 2007 to 2011. Then Percolate, where he was recruited out of the consulting arm onto the product team and pushed a multi-year digital asset management project across the finish line - the kind of project that teaches you what it takes to change a system that a lot of other teams depend on. Then Frame.io, the video review platform Adobe would later buy for a billion and a quarter.

At Frame.io he found Chris Bell, now Knock's CTO, and the outline of an idea. Every product team they knew was building the same notification pipeline from scratch and paying for it every quarter in tech debt and paged engineers. It was not a market failure so much as a market gap. Nobody had built the thing.

So in 2021 they did. Knock launched publicly with the language most infrastructure companies use only after their Series B - talking about workflow engines and observability, not just delivery. Seely describes the product as "API first, and a dashboard second," which is a five-word GTM strategy: obsess over SDKs and API documentation, and let the developer's evaluation experience carry the pitch.

The Elixir bet

Knock is built on Elixir, which is a slightly unusual choice for a company hiring aggressively in 2026. The language runs on the Erlang VM and is quietly excellent at exactly the kind of concurrent, long-running, message-passing work that a notification workflow engine has to do. It is also a hiring filter - engineers who choose Elixir tend to be the sort who read papers on distributed systems for fun. Which, given the pitch about where the best notifications engineers should go work, is probably the point.

The cold email that stuck

In a LinkedIn post from late 2024, Seely wrote that in Knock's earliest days he sent a cold email that turned into a customer, then a case study, then part of the story he tells about how the company got off the ground. It is the sort of anecdote founders share to remind themselves that boring habits build interesting outcomes. It is also, on his part, a mild flex: the person who runs an API company for engineers still writes personal outbound.

His personal site is one page. There is a dithered black-and-white portrait, a paragraph about him, a link to his Twitter, and - unusually - a public Notion page of every book he has been reading, plus a note that he is always taking recommendations. He is 28 employees deep into running a Series A company, and one of the first things he wants strangers to know is what he's been reading.

On the "Stripe of X" comparison

Every infrastructure founder eventually gets asked whether they are the Stripe of something. Seely's version of the answer is the payments-vs-search-vs-notifications framing above, which does two useful things at once. It anchors Knock to a well-understood category shape - developer-first infra with clean docs and a workflow engine - and it makes the argument that notifications belong in the same tier of importance as payments and search, which is not something a customer would have said unprompted in 2020.