Introducing HQ: intelligence on a handle

Most AI tools stop at the conversation. They answer, summarize, and suggest, then hand the work back to you. HQ starts where they stop. You @mention a specialist in Slack or Teams, and it researches, decides, and does the work, then reports back. We call it intelligence on a handle, and it is what we set out to build at The Intelligence Company.

A team of specialists, not one generic assistant

You do not get a single chatbot. You assemble a roster of specialists, each with its own model, skills, integrations, and tone, and each reachable by an @handle. Mention @hq dev for engineering, @hq scout for research, @hq nova for data. The right specialist picks up, answers in its own model and voice, and the thread stays with it from the request to done.

A real machine behind every handle

The difference between chatting and doing is a computer. Every conversation in HQ runs on its own private, hardware-isolated machine, the same class of isolation behind modern serverless platforms. It boots in a fraction of a second and resumes exactly where it left off, even days later. Because it is a real machine, a specialist can run a script, crunch a file, drive a browser, build a small tool, and install whatever it needs, instead of only telling you how.

That machine is qOS, the operating system we built for agents, in Rust and top to bottom: search, cache, storage, graph, browser, email, documents, extraction, memory, and the model layer. We did not staple it together from other people’s APIs. We built the data layer ourselves.

Research you can act on

A decision is only as good as what it rests on. HQ specialists browse the live public web the way a person does, from the country you choose, without being blocked, throttled, or served a stripped-down page. They reach sources, regions, and logged-in tools an ordinary assistant cannot, and hand back an answer with its sources, not a confident guess.

The right model for each job, including ones we run ourselves

Decision intelligence needs more than one model. HQ routes every call across the frontier, Anthropic, OpenAI, xAI, and Cerebras, and sends each task to the model that wins on quality, latency, or cost. That frontier includes Claude Fable 5, Anthropic’s most capable model, and we add new ones the day they ship. Underneath, the high-volume and data-sensitive work, the embeddings and reranking behind memory and search, runs on open models we serve on our own GPUs, so your most sensitive text never leaves our infrastructure to be turned into numbers.

Trust is the foundation, not a setting

Anything that can act needs a trust model that holds. Every action a specialist takes is bound to the person who asked and the permission they were granted, checked at each step, and written to a log that cannot be edited after the fact. You choose whether your data lives in the EU or the US. We are based in Stockholm and built HQ under GDPR and the EU AI Act from the first line, not bolted on after. You can always see what a specialist knows, and correct, delete, or freeze it.

Where we are

HQ is invite-only while we onboard teams carefully. Slack and the browser extension are live today, with Microsoft Teams on the way. If you want a team of specialists in your own workspace, request access.

This is the first of many notes from the team. We will go deep on the sandbox, the identity chain, memory, and the model layer in the posts to come.