Like an operating system, only for AI
The cloud separated infrastructure, platform and applications (IaaS, PaaS, SaaS) and made every layer replaceable. The AI operating system follows the same principle – with one difference: all layers run in your own tenant, not with the provider.
| Operating system | AI operating system | At innFactory |
|---|---|---|
| Users | People and AI agents | Employees from apprentices to management, domain and coding agents |
| Applications | Work surfaces | CompanyGPT, Office add-in, Claude Code, Cowork, Codex |
| Libraries and services | Company harness | Knowledge, skills, methods, processes, roles |
| Kernel | Control plane | innFactory AI Gateway |
| Processors | AI models | Claude, GPT, Gemini, Mistral, open models |
| Hardware and drivers | Core systems and data | ERP, CRM, DMS, Microsoft 365 in your own cloud tenant |
CompanyGPT and AI Gateway: together or separately
Both products are built to work together seamlessly, but each also does its job on its own. Three typical starting points:
- CompanyGPT only: you equip every employee with a GDPR-compliant AI workplace, with knowledge base, agents, integrations and Office add-in. For many companies this is the first step.
- AI Gateway only: your development teams already use Claude Code, Codex or custom agents. The gateway puts cost centres, budgets, SSO and guardrails in front of all access without replacing the tools.
- Both as the AI operating system: CompanyGPT runs through the gateway. One set of rules, one budget and one analytics view cover chat, Office, coding agents and domain agents.
One budget for every AI tool
In the AI operating system every person gets a budget, not every tool. Whether someone works in the CompanyGPT chat, creates a summary in Word or refactors a function in Claude Code, everything books to the same cost centre. Teams get hard limits, and management sees in one analytics view which department uses which models and what that costs. For tools with their own subscription, the gateway’s companion measures agent time per repository and project so that this usage becomes transparent as well.
Operation, procurement and costs
The platform runs in your cloud tenant on Microsoft Azure, AWS, Google Cloud or STACKIT. innFactory sets it up, trains your teams and takes care of maintenance on request. If you like, you source the cloud resources and model access directly through innFactory GmbH as a cloud reseller: by invoice instead of credit card, with all providers on one invoice.
| Building block | Price |
|---|---|
| CompanyGPT | from EUR 14,990 fixed-price package, no per-seat fees |
| AI Gateway | from EUR 199 per month, tiered by user count |
| AI models | token costs directly with the provider, no margin for innFactory |
All prices plus VAT.
Your path to the AI operating system
- Clarify the strategy: which processes should AI take over, which data and systems are involved? We work this out in our AI strategy consulting.
- Equip every employee: CompanyGPT as the shared AI workplace, accompanied by AI training.
- Steer the agents: put the AI Gateway in front of coding agents, pipelines and domain agents, with budgets, guardrails and SSO.
- Build the harness: describe knowledge, skills and processes so that people and agents can use them together.
- Hand over core processes: bring agents into core processes step by step, connected via MCP and monitored in the gateway.
For professional secrecy holders such as medical practices, law firms and tax advisers, for industrial companies with sensitive design and production data and for the public sector, operating the platform themselves is often the precondition for using AI in core processes at all.