Inbox under control
Your assistant reviews e-mail several times a day, sorts by urgency and prepares draft replies. You decide, it types.
One assistant per person.
One system for the company.
Every person on the team gets a personal AI assistant – with memory, access to mail, calendar and documents, available where the work happens. Not another subscription. A system you own.
The AI system deployed by Staub IT is open and LLM-independent, built on proven open-source components such as Hermes. It combines personal AI assistants with a shared company memory: every person works with their own assistant that knows their projects, appointments and documents – while the company keeps knowledge, data and control in-house.
What your assistant takes off your plate.
No future tense here – everything below runs in production today, for us and for our clients.
The day starts prepared
Morning briefings with appointments, open items and context for every meeting – before you finish your first coffee.
Knowledge on demand
Contracts, project status, decisions from three months ago: your assistant finds the answer in your documents and cites the source.
Routine runs itself
CRM reconciliation, accounting workflows, recurring reports – defined processes run reliably in the background.
AI answers the phone
An AI phone assistant with a real Swiss number takes calls, answers standard questions and forwards whatever needs human attention.
Available where you work
Chat, e-mail or phone – the system fits into your environment, whether Google Workspace or Microsoft 365.
The foundation
Built to last.
AI models change every few months. The system is built so your company benefits from that, instead of starting over each time.
Open, no lock-in
The system is built on open software and runs on your own infrastructure – hosted in Switzerland if you wish. You can inspect, take along or extend everything, at any time.
LLM-independent
It works with every major language model. When a better one appears, we swap it in – no migration, no new project. Your system stays state of the art.
Memory by design
A shared company brain for facts, projects and decisions, plus a personal memory per person. Your company gets smarter with every day of use.
Clean permission boundaries
Every person sees only what they are allowed to see. Assistants inherit their users' permissions – confidential stays confidential, including from the AI.
From pilot to team system.
We introduce the system in stages – keeping risk small and value visible from day one.
Pilot with a defined goal
A few people, one workflow with immediate value. At kickoff we put in writing what the system must deliver.
Expansion into the team
Person by person, with training (two to four hours each) and fine-tuning in live operation. Stages instead of big bang.
Operations & evolution
Updates that preserve your customisations, monitoring, support and ongoing model upgrades. New workflows join when they pay off.
You pay for an outcome, not for effort: what we agree at kickoff, we deliver – and we keep working until it runs.
What decision-makers ask us first.
Where does our data live?
On your own infrastructure – on a server in Switzerland if you wish. The system is not a third-party cloud platform: you always know where your data is and who has access.
Which AI model is behind it?
Whichever is best at the time. The system is LLM-independent and works with the major models from Anthropic, OpenAI, Google and others – we pick per task what fits best, and switch as soon as something better appears.
What if Staub IT is unavailable?
Nothing that locks you in. The system consists of open software and runs on your infrastructure. The documentation is yours; any capable IT firm could take over operations. We want to convince you, not chain you.
How long does a rollout take?
A pilot with a few people is up in a matter of weeks. Expansion into the team follows in stages – your processes set the pace, not our calendar.
What does it cost?
That depends on team size and scope – any serious answer starts with a conversation about your processes. The entry point is deliberately small: pilot first, expansion after.
See the system on your own processes.
In an intro call we show, based on your workflows, what an assistant could take over tomorrow – concrete instead of abstract.