Let AI agents do the work. Your team approves it.
Put an agent on any step with plain instructions in Arabic or English. It reads the run's data and your documents, returns a structured answer, and anything important waits for a person.
You choose what it can use
Your connections, your knowledge base and the data of the run. Nothing else.
Your model, if you want
A monthly platform AI allowance is included. Any plan can connect its own OpenAI-compatible model.
Approval before it acts
Payments, posts and sensitive steps wait for a named approver.
How you decide what an agent can do.
Plain instructions, structured results
Write what the agent should do the way you would tell a colleague. The answer comes back as fields the next steps can use.
Your knowledge
Knowledge search answers from your own documents, in Arabic or English.
Reads documents
Pulls fields out of invoices, PDFs and scans.
Human approvals
The run pauses, the approver decides, and the run continues from there.
Your own model
Connect any OpenAI-compatible endpoint, including a self-hosted one, on any plan.
Flow Scorecard
Write a rubric, and a judge model scores each run's AI output from 0 to 100.
AI Desks Enterprise
Ready teams of agents for research, writing, sales and operations: specialists run in parallel, a coordinator writes and scores one deliverable, and a person signs off.
Common questions
What does an agent step come with?
Instructions, the data of the run, optional knowledge search and document reading, and a structured result the next steps use.
Can we use our own AI model?
Yes, on any plan: any OpenAI-compatible endpoint, including self-hosted models through Ollama, vLLM or LocalAI.
How is platform AI charged?
Each plan includes a monthly platform AI allowance. When it is used up you can top up, or connect your own model.
Which plan includes AI Desks?
AI Desks are part of the Enterprise plan. AI agent steps are available on every plan.

