Assistants that know your company.And reach your systems.
An agent is skill, model, tools and knowledge in one package — started in chat, shared with the team, connected to the systems your admins approve.
We are launching soon — secure one of the first spots.
Three agents stacked, the middle one in focus: the quote assistant with file search, graph and three knowledge files.
What makes an agent
Not a prompt note, but a definition the team reuses.
Skill
The instruction with variables — how the agent works and what it asks first.
Model & mode
Fixed or auto mode; by default always an EU model.
Tools
Web search, file search, company graph, image generation and approved connections.
Knowledge
Referenced files from the knowledge system — no second upload path, no copies.
Connections to your systems
A curated catalogue, grouped by seat and place of processing — plus your own servers via open standards.
Catalogue, EU and US apart
Vendors with an EU seat or self-hosted software come first; US vendors are clearly marked — including where processing happens.
The platform's connector catalogue: three groups — EU & self-hosted, US vendors with EU processing, US vendors — with categories such as CRM, tickets and Git hosting as chips; the EU group comes first, the US group is greyed out.
Open standards
MCP servers and REST interfaces, with personal sign-in per person — your rights in the target system remain your rights.
Allow-list instead of blind flight
Admins define per connection which tools are allowed; writing tools need explicit approval.
Foreign content stays foreign
What comes back from tools counts as untrusted and passes a deterministic check layer — no tool can redirect the model.
Governance that convinces works council and IT
Who may do what — and who checked it? Both live in the platform, not in a spreadsheet beside it.
AI usage policy
One policy per organisation that every person acknowledges — versioned, with proof.
Approval lifecycle
Agents and skills are unreviewed, approved, flagged or disabled — with reason and audit entry.
Inventory & register
All agents and skills at a glance, with model, tools and approval — the basis for the AI system register.
Inventory of agents with approval status: quote assistant approved, research assistant unreviewed, termination helper flagged; below it the usage policy with 24 of 26 acknowledgements.
Shared like everything else
Private on creation, then deliberately shared: with groups, individuals or the whole organisation — to view, use or edit.
- Start from the agents page or right in the composer
- Share with groups, people or organisation-wide
- Chats with an agent stay personal
Agents come with the seat
Agents, tools and connections are included in every platform seat. A Work slot is only needed once an agent should run on a trigger.
€29 per user/month
Usage: standard€79 per user/month
Usage: approx. 3× standard€199 per user/month
Usage: approx. 8× standardIncluded in every seat — there is no feature gate. The tiers differ only in the monthly usage allowance.
All prices net per seat and month, plus 19 % VAT.
More building blocks of the AI infrastructure
One login, one permission model, the same sovereign infrastructure.
Ready to bring your AI infrastructure to Europe?
Up and running in minutes — pick a tier, cancel monthly.
We are launching soon — secure one of the first spots.