Bootstrapped Private GenAI Startup Hits $1M Annual Revenue, Launches Helix 2.0
The people behind the story, how agentic AI is changing and why we don't want a sales call with you
It was late 2023 when we decided to do startup #3. This time, having experienced first-hand the impact that the “ChatGPT moment” had on my consulting clients, I saw the opportunity for a private GenAI stack that you could run on your own infrastructure.
I was fortunate to have saved enough from consulting so that I was able to stop earning for a year and dive headfirst into building product again. Phil and Chris did the same, and we committed to bootstrapping this thing. (If you haven’t read “Reconsider”, I recommend it.)
Since then, we’ve had our share of ups and downs – but having intentionally not taken any external funding, I’m pleased – and significantly relieved – to be able to announce that we’ve just hit the milestone of $1M in annual enterprise revenue.
I can’t speak publicly about who the customers are yet, but suffice to say we are lucky to have some of the most innovative hedge funds, investment managers, and service providers in the world working with us.
Why are they investing in AI Agents on a Private GenAI Stack?
Because AI Agents will (actually) change everything
In the early days of Helix, I had a healthy degree of skepticism that AI might be more hype than substance. But it has been this year, 2025, that has convinced me that we’re on a trajectory for AI Agents to be a true new industrial revolution. What convinced me? It was using the agents in my own work. If you look at Cursor with Claude 4, it’s made me 30-40x more productive than I used to be armed merely with vim. If you look at the hours worth of research you can do with Perplexity in minutes, I’m able to make informed business and technical decisions in a fraction of time it took before.
So is that it? We all hand our data over to these AI SaaS companies, and they take the whole pie? Wait up – there’s a problem lurking in big businesses. With increasing geopolitical strife and threats from security breaches, enterprises are ever-more sensitive to where they send their data, and how they protect their core IP. Turns out, lots of companies are not comfortable sending a lot of their data to OpenAI, or even Microsoft. Private cloud is back. So where’s the private cloud GenAI stack?
Enter Helix 2.0: The Fastest Path to AI Agents on a Private GenAI Stack
You can read more about it on our shiny new website, but the goal here is to be the Macbook Pro of GenAI stacks.
Ever tried using Linux on the desktop? Spent time recompiling your kernel so you can get your webcam working? Wonder why people buy Apple products even though they’re expensive? It’s because they just work, from soup to nuts. That’s the goal with our GenAI stack for running on your own infrastructure.
We give you everything you need to run AI Agents, connected to your data and business systems, either on open source models like Qwen3 running on your own GPUs or proprietary LLMs that you can provision in your VPC, like Claude 4.
Here’s a demo of our latest stuff, check it out:
Wait, you don’t want a sales call with me?
So sometime last year I was talking with my friend John Merrells and he made the excellent point that most people under 40 don’t want a sales call to be able to buy something. So we put tons of effort into making Helix easy to self-serve, with transparent pricing.
It’s not that we don’t want to talk to you – we talk to our customers all the time, pair with them, fly to their offices to run workshops and co-develop agents with them – but instead of being forced to sit through a slide deck, you can deploy Helix yourself in minutes, in Docker, Kubernetes or any major cloud. If you have a question then just hit the chat box in the bottom right of the website (it will connect you to the real team, not an AI). You can evaluate Helix yourself and provision a license through our new self-service Launchpad system where you can deploy:
Agents onto Helix Cloud, our SaaS demo environment - where you get a regular user account
Trial VMs - which give you root access to a VM and full admin access to Helix, although they are configured to talk to external inference providers so are not fully private. We spin these up for you instantly, because we keep a warm pool of them for instant use
GPU Instances - we offer 2x A100 80GB GPU nodes at just $5/hour through our partner Civo, who have been awesome to work with
Everything you can do with Helix, from multi-turn agents integrated with business apps and vision RAG over complex document layouts, can all run on a single A100 GPU, fully private.
Cheers to $1M revenue!
So cheers, here’s to the team who made this happen (including our secret co-founder), folks joining and helping out, the customers who put their trust in us, and all of you crazy AI labs out there building SOTA LLMs – thank you!
Check out helix.ml to build AI agents on Helix, deploy to your own infrastructure, and eliminate tedious work in your business.
For a slightly more formal take on the 2.0 release, and more information, check out the press release: Helix 2.0 Gives Global Enterprises the Fastest Path to AI Agents on a Private GenAI Stack
Congrats!