p6k is built by two engineers who were in the room when the last great enterprise platform was written — and who have spent the twenty years since learning what the next one has to get right.
Bow came up at Peregrine Systems in the early 2000s — administering and then extending the asset and service management software that a generation of IT departments ran on — and learned the trade the way it is actually learned: by shipping software that other people depend on, and being on the hook when it breaks.
In 2005 he joined a two-person company called Glidesoft, working on a service management product almost nobody had heard of. It became ServiceNow. Bow was one of the original architects of the Glide platform — the record model, the scripting layer, the UI engine that everything above it was built from — and he stayed as Principal Architect while the company went from three people to thousands, and on to an IPO. Code he wrote in those years is still running inside a large share of the Fortune 500.
In 2013 he left to find out what he would do differently with a blank page. That became Dreamtsoft, an enterprise-grade, multi-tenant, zero-trust application platform he founded and architected: schema-less Postgres underneath, plain HTML, JS and CSS on top, Git as a first-class citizen so upgrades and team development stopped being a special event. It was a deliberate answer to every architectural decision he had watched calcify at scale.
Most recently he was at StarLifter, Fred Luddy’s new company, working alongside Jay again — the same people, a smaller table, twenty years after the first one. He left at the end of March 2026.
P6K is the culmination of everything Bow has spent his career building toward. His belief is simple: AI is already capable of doing far more work than most businesses can actually hand over to it. The problem isn’t the intelligence of the models — it’s that the business itself is scattered across systems, processes, tribal knowledge, and undocumented decisions. P6K gives that complexity structure. It turns the way a company actually operates into something AI can understand, reason about, and act on. Building that foundation is a platform problem — and platform problems are where Bow has spent his career.
Jay arrived at the same three-person company by a side door. In June 2005 he was contracting through JMI for Glidesoft — the company that would rename itself ServiceNow — and he simply never left. He was one of the small handful of engineers who wrote the original platform, back when, as he has put it, there were only about ten people and the whole thing was about getting together and doing something that mattered.
From 2007 to 2013 he was Chief Architect at ServiceNow, which is the job of holding a platform’s design together while everything around it changes: a startup becoming a public company, a handful of customers becoming thousands, a product becoming a category. He carried the architecture through the 2012 IPO and the hypergrowth on either side of it, then took on Quality Engineering and Productivity as Senior Director — because at that size, how fast and how safely the engineering organization can move is an architecture problem.
He went back in 2020 as a Distinguished Engineer, which gave him something very few people have: a close look at the same platform he helped design, twenty years and a hundred billion dollars of market capitalization later. What still held. What had to be worked around. What everyone had quietly agreed to live with. He has also lent his experience to others by serving on the board of RizePoint for several years, culminating in an acquisition of the company.
In late 2022 he became CTO of StarLifter, Fred Luddy’s new company, where he and Bow worked together again until the end of March 2026.
At p6k he is the person asking the unglamorous questions — about upgrade paths, tenancy, failure modes, and who is accountable when an agent is wrong — which is exactly the set of questions that separates an enterprise platform from a demo. He has watched what happens when they are answered late.
Bow and Jay have been building the same kind of thing, next to each other, on and off since 2005: the original Glide platform, then StarLifter — the last stretch of it spent working out of Fred Luddy’s house, twenty years after the first time — and now p6k. That is an unusual amount of shared vocabulary. It means arguments get to the actual disagreement quickly, and it means neither of us has any illusions left about how much of a platform is the part nobody sees.
It also means we have watched the same mistake made repeatedly, by good teams: shipping the surface first and hoping the system underneath can be added later. It cannot. That is why we have spent longer thinking through p6k than a faster team would, and why the parts we are proudest of are the ones you will never look at.
There is a reasonable chance the person who reads your note is the person who built the thing.
Contact us