Skip to content
oneAI
Log inBook a demo
oneAI
Book a demo
Log in

oneAI

Chat, knowledge and agents. Under your company's rules.

The workspace your teams use every day: every frontier model, the knowledge in your systems and agents that take on routine work, with redaction, approvals and an audit trail built in.

Book a demoStart for free
Colleagues at the long oak table of the pavilion in the morning sun, one at her laptop, one leaning in with a printed document, the lake and the Alps behind the glass.
Seven vendors, one picker
Each model says where it runs, and admins decide which ones the company sees.
Connected to your systems
SharePoint, Outlook, weclapp, Business Central, HubSpot and more, with over 200 tools.
Governed from the first prompt
Rules redact or block before a model sees the text, and every request is logged.

What it does

Everything your teams do with AI, in one place. And everything your admins need to allow it.

  • Every frontier model

    Pick a model per chat or let Auto choose per message. Each shows where it runs and what a message costs.

  • Work that leaves the chat

    Documents with tracked changes, presentations in your PowerPoint master, apps and dashboards, edited together live.

  • Answers from your own knowledge

    Keyword and meaning search with a reranker across 89 file types; every answer shows its sources and keeps their rights.

  • Agents with an approval queue

    Build an agent in four steps, run it on a schedule or when something happens in a system, and park every write for a person.

  • Redacted before it leaves

    23 built-in rules for personal data, IBANs, secrets and EU AI Act categories. Each can log, flag for review, block or redact.

  • MCP both ways, and an API

    Use oneAI from Claude, Cursor or VS Code, connect any MCP server to it, and call it like OpenAI or Anthropic.

In daily work

Less copy and paste. More finished work. Three moments from an ordinary week.

  • Compare the three offers for the new CNC machine.

    Kessler is cheapest over five years. Brandt delivers six weeks sooner and includes training.

    • Angebot-Kessler.pdf
    • Angebot-Brandt.pdf
    • Mail from Huber
    Ask anything across your systems

    Answers across your files

    Ask across documents, mail and systems at once; the answer names every file it used.

  • Supplier portal is live10:14

    On lieferanten.nordwerk.de, certificate issued.

    Three new submissions11:30

    Written to the table Suppliers.

    Apps that go live

    An app built in a chat publishes on your own domain, and its forms write straight into your tables.

  • claude_desktop_config.json
    "mcpServers": {  "oneai": {    "url":      "https://hub.oneai.eu/api/mcp",    "headers": {      "Authorization": "Bearer <API key>"    }  }}

    oneAI inside your tools

    Claude, Cursor and VS Code reach your Spaces and projects through oneAI's MCP server, under the same rules as the chat.

What a chat leaves behind

Not just answers. Finished work. Built in the chat, edited together, kept in your Spaces.

Rahmenvertrag-Kessler.docx

§ 7 Liability

7.1 The supplier is liable for damage caused by defective deliveries in accordance with the statutory provisions.

7.2 Liability for slight negligence is limited to the order valueforeseeable, typical damage, except for injury to life, body or health.

Checked against your purchasing templateAcceptReject

oneAI suggests changes as tracked changes against your own templates; colleagues edit the same document live.

Ecosystem

Two products plug straight in. Each runs on its own, and oneAI uses them natively.

oneAIChat, knowledge, agents
oneStackOpen-weight models on your GPUs.
oneDeployEU cloud or on-premise, kept current.
  1. Private model · EU GPUs01
  2. Custom domain · certificate02
  3. Private model · EU GPUs01

Everything in oneAI

One workspace, governed end to end. In English and German, in the EU cloud or on your servers.

Chat

  • Models from Anthropic, OpenAI, Google, Mistral, DeepSeek, Qwen and xAI
  • Where each model runs, and what each message costs
  • Auto picks a model per message; reasoning effort per chat
  • Branch a chat, or answer again with another model
  • Mention files and Spaces, follow-up suggestions
  • Rewrite in another tone, distil a chat into a brief or a spec
  • Voice mode you can interrupt, with spoken approvals
  • 14 image models for creating and editing images
  • Web search, a research agent and a browser the agent can use

Knowledge

  • Spaces with their own access rights
  • 89 file types, images and scanned pages read by a vision model
  • Keyword and meaning search with a reranker
  • Git repositories indexed with their code structure
  • Data sets as tables, queried with SQL
  • Projects with instructions and 50 templates, German law included
  • Persona store and personal memory

Work products

  • Documents with tracked changes, co-edited live
  • Presentations in your PowerPoint master, PPTX and PDF export
  • Apps in a sandbox, published on your own domain
  • Dashboards over your data
  • Python and SQL run for analysis

Agents and automation

  • Agent Builder in four steps: model, tools, knowledge, rules
  • Schedules and triggers on events in connected systems
  • Approval queue for every write an agent wants to make
  • n8n workflows built and edited from the chat
  • On-premise agent that reaches local file shares
  • Uno, the built-in guide that sets things up for you

Integrations and API

  • SharePoint, OneDrive, Google Drive and network shares
  • Outlook mail and calendar, HubSpot, Dynamics 365 Sales
  • Business Central, weclapp and Lexware Office
  • GitHub, GitLab, Forgejo, ClickUp, Fireflies, NinjaOne, Plytix
  • n8n, external APIs and any MCP server
  • oneAI as an MCP server for Claude, Cursor and VS Code
  • OpenAI- and Anthropic-compatible API under the same rules

Governance and administration

  • 23 built-in rules: log, review, block or redact
  • Compliance model rating EU AI Act risk and data categories
  • Every write confirmed, by card, policy or a time-boxed trust window
  • Audit log with export and a review queue
  • Roles, teams with their own models, seat types
  • Sign-in with Microsoft, Google, Apple, passkeys or TOTP
  • SCIM provisioning of users and groups
  • Budgets, credit pools and quotas per person

Questions

What teams ask before they start.

Another question?Talk to us

Models from Anthropic, OpenAI, Google, Mistral, DeepSeek, Qwen and xAI, plus your own through oneStack or any compatible endpoint. Each model says whether it runs in the EU, and admins decide which ones the organisation sees.

The oneAI family

Three building blocks, one sovereign AI OS. Sovereignty is a matter of freedom of choice — we offer the choice.

oneAI, oneStack and oneDeploy run in the cloud, in your private cloud or on-premise. Each works on its own, and all of them work together natively.

Talk to us

  • oneStack

    Open-weight models on your GPUs.

  • oneDeploy

    EU cloud or on-premise, kept current.

Next step

See oneAI inside your company.

Thirty minutes with a solutions engineer, on your systems and your use cases.

Talk to a partner

Book a demo

A person from oneAI answers within one business day. No ticket queue. By sending the form you agree that oneAI stores and processes your details to answer your request. Privacy policy