Pitches · Pre-seed · November 2026
Tabs & Spaces by ProteusAI
A coworking space for people and AI agents, for companies that have to own their data.
Think Grok Bot or OpenAI Dots: agents with computers of their own, doing real work beside your team. But on infrastructure the company controls, with the models it chooses, so banks, hospitals, law firms and insurers can finally use AI.
01 / 11
02 The problem
The companies that would gain most from AI can't use it.
Every question asked of an AI tool, and every document given to it, leaves the company: to the AI company's servers, its models and its staff. For most businesses that's a trade-off. For a regulated one, it's out of the question.
So their people keep doing by hand what agents could prepare in minutes, or they paste customers' data into tools they shouldn't. In the AI era, owning your data has gone from a nice-to-have to the condition for using AI at all.
03 Why now
Agents that do real work arrived this year, in someone else's cloud.
In August, xAI launched Grok Bot: agents with their own computer, signing into the tools you use and working while you're away.1 In September, OpenAI launched Dots: always-on agents with their own cloud computers, working across the apps you connect.2
Both prove what agents can now do. Both run on their maker's cloud, with their maker's models, so they're out of reach for the companies that need help most.
- Regulators are watching AI closely. The EU AI Act's obligations are phasing in, and creditworthiness checks on people are high-risk under it.
- Open models are good enough to do real work on a company's own servers.
- Companies have the infrastructure: their own data centres and private clouds, and the people who run them.
04 The solution
Tabs & Spaces: people and agents, working together, on the company's own infrastructure.
A workspace where teams and their agents work side by side: a space for every team and topic, with tabs for conversations, files, tasks and apps. Agents have instructions, computers of their own and a budget; they answer, carry out tasks on a schedule, and ask before they change anything.
On infrastructure the company controls
Run it yourself on your own servers, or on ours, with your own database and bucket. Nothing reaches us unless you send it, and every action, by people and agents, is in an audit log that shows if it's been changed.
05 What's live
Built, running, and used by paying customers.
06 Example: a bank
A bank's teams, each with an agent beside them.
We've made a ready-made workspace for a bank: start from the template and it's set up in seconds, with its spaces, agents, apps, routines and example policies.
People decide, with four eyes; agents never decide on a customer or a credit. See how it works →
07 Traction
Early, honest numbers, and Dutch banks next.
With a Dutch partner, a startup still in stealth, we're already building AI solutions for Dutch banks. Banks are exactly who Tabs & Spaces is for, and we expect to be big in that space (pun intended).
08 Business model
Subscriptions in the cloud, support for those who run it themselves.
Regulated companies start on their own infrastructure, with us supporting them; teams that can use our cloud start there in minutes.
09 How we're different
The agents of Grok Bot and Dots, the ownership of software you run yourself.
| Grok Bot, OpenAI Dots | Tabs & Spaces | |
|---|---|---|
| Where data and agents run | The vendor's cloud | Infrastructure the company controls |
| Models | The vendor's own | Any: open models on its own servers, or a provider it chooses |
| Run it yourself | No | Yes, or in our cloud with your own database and bucket |
The agents' runtime is OpenClaw, an open-source project. We maintain our own fork, so what agents can do on a company's servers is in our hands and theirs.
10 Team
We build Tabs & Spaces with Tabs & Spaces.
Before writing a line of it, I spent two years speaking with companies and founders about how AI could help them. One blocker came up again and again: their data couldn't leave. Regulation and privacy laws kept the companies with the most to gain from using AI at all.
I started ProteusAI full-time in January 2025 to fix that, and built Tabs & Spaces to run services and language models on infrastructure companies own or control.
We were a team of seven; financial constraints brought us down to two. Agents on Tabs & Spaces keep us shipping as if we were more: they draft, review, research and test beside us every day.
ProteusAI · registered in the Netherlands in 2024 · operating since January 2025 · tabsandspaces.dev · developer.tabsandspaces.dev
11 The ask
€400,000, pre-seed, to put Tabs & Spaces inside regulated companies.
We haven't raised before. This round pays for three things:
- Running the company while the first banks and their partners come on board.
- Engineering to make installing it simple on a company's own infrastructure, its models included, so its own engineers can do it without us.
- Certifications regulated buyers ask for: ISO 27001, SOC 2 and SOC 3.
Tabs