I'm a solo founder, engineer, cyclist, photographer and unapologetic cat person. I've been in IT since 2018, but I've never been particularly good at staying inside one box.
I've worked on backend systems, automation, infrastructure, AI, microcontrollers, TinyML, blockchains, and game development. I've built things that were useful, things that were weird, and a lot of things simply because I wanted to see if I could.
Since 2022, my wife, our cat and I have been travelling around the world. With the world getting a little crazier every year, we decided to see more of it while we could. We've lived in Southeast Asia, Central Asia, and now we're in Georgia.
At some point, building things stopped being something I did on the side.
It became the job.
Over the last year I've been working on AI systems as a freelance engineer. I've built RAG systems, multi-agent workflows, support systems, document automation and integrations with external platforms.
And eventually I kept running into the same problem. AI agents are getting very good at deciding what to do. I'm much more interested in what happens when they actually do it.
- An agent can send an email.
- It can issue a refund.
- It can modify customer data.
- It can delete something.
The model might have a good reason. It might have a bad reason. It might be completely wrong. Either way, something still has to decide whether that action is allowed.
That became the idea behind NullRun, the company and product I'm building now.
NullRun sits between an AI agent and its tools. Before an action executes, it checks the declared policy against the actual action and returns allow, block, or require_approval.
I'm building the whole thing myself: architecture, Rust backend, Python SDK, infrastructure, product and the parts of the website I probably shouldn't be spending this much time on.
The product is still early. No customers yet. No team. No funding.
Just me, a codebase, and 91 names in a spreadsheet that I'm slowly trying to turn into conversations.
That's also what I want to write about here.
I'll write about building a company as a technical founder, making product decisions without a big team behind you, engineering systems that have to behave predictably, and the things I learn when ideas meet actual users.
I'll also write about the technical side of agent execution, security and runtime authorization when there's something worth digging into.
And occasionally I'll probably write about cycling, photography, travelling, or some completely unnecessary thing I decided to build.
That's what I'm doing these days.
Top comments (0)