Works with the AI model you choose
Your business.
Your AI workforce.
Create, connect and manage AI agents that handle everyday business tasks. No coding required.
Includes an offline sandbox model, so you can try every feature before adding an API key.
How AgentOS works
- 01
Describe your business
Tell AgentOS what you do and what you want help with, in plain English.
- 02
Pick agents from the marketplace
Ready-made agents for reception, quotations, follow-ups, support and more.
- 03
Drag them into a workflow
Connect agents, approvals and your apps on a visual canvas.
- 04
Approve what matters
Emails, quotes and payments wait for a person. Every action is logged.
Popular agents
See all 16 templates →Receptionist
Administration
Reads enquiries and routes them
Quotation Agent
Sales
Drafts quotes from your catalogue
Finance Follow-up
Finance
Prepares payment reminders
Support Triage
Customer Support
Classifies tickets and drafts replies
Project Coordinator
Construction
Turns jobs into task lists
Quality Reporter
Manufacturing
Structures inspection reports
Visual workflow creation
Drag agents, conditions, approvals and app actions onto a canvas. Validation explains problems in plain English.
Human approval and control
Approvals are bound to the exact payload. If the action changes, it needs fresh approval. Agents can never approve themselves.
See what agents do
Every step, input, output, token and cost is recorded. Budgets stop spending before limits are exceeded.
One platform, many industries
Restaurants
Receptionist, reservation coordinator and inventory agent.
Manufacturing
Quality reporting, production documentation and purchasing drafts.
Legal offices
Document intake and contract organisation with mandatory professional review.
Construction
Enquiries, quotations, job organisation and payment reminders.
E-commerce
Order enquiries, returns and catalogue updates.
Healthcare admin
Non-clinical scheduling and administrative intake.
Integrations
Available today: any SMTP mailbox, webhooks and form submissions, generic REST APIs, a built-in CRM and document knowledge bases. AI models from any major provider or your own servers.
Security by design
Tenant isolation in the database, credentials kept in a server-side vault, a server-side tool gateway and append-only audit logs. MFA and GDPR request workflows are on the roadmap.
Read the security overview →- Permissions are enforced on the server, not by the AI
- Retrieved documents are treated as data, never as instructions
- Duplicate emails and payments are prevented by idempotency keys
- Budgets and rate limits stop runaway spending
Frequently asked questions
Do I need to know how to code?
No. You describe what you need, choose agents and connect them visually. Advanced users can still use webhooks, APIs and JSON schemas.
Which AI models can I use?
Bring your own: OpenAI, Anthropic, Google Gemini, Azure OpenAI, Mistral, Groq, DeepSeek, OpenRouter or a self-hosted model such as Ollama or vLLM. You can mix providers per agent.
Can an agent send an email on its own?
Not unless you allow it. Sending email, calling external APIs and similar actions require a human approval bound to the exact message, enforced on the server.
Where is my data?
Each organisation is isolated at the database level. API keys are stored in an encrypted vault and never shown to the browser or the AI model.
Is AgentOS compliant with the EU AI Act?
AgentOS includes a structured AI Act self-assessment and GDPR tools, but a checklist is not a legal determination. High-risk uses require qualified legal review.
Ready to give your first agent a job?
Start building