An organization of AI agents that works for your company.
The organization stays yours, even when the AI changes.
An org chart for people and agents.
Decide where each agent belongs, who is responsible for it, and which people and agents it works with. We help you build and manage that organization across your computers and teams.
Each agent gets an identity your company controls, with a person responsible for its work.
Your company
Operations team
The operations lead is responsible for these agents.
- ContractsAgent on your computer
- PurchasingAgent on a company server
- ReviewAgent on a colleague’s computer
Client team
The client lead is responsible for these agents.
- ResearchAgent on a company server
- DraftingAgent on the client lead’s computer
Teams set what each agent can reach. Keep clients, partners or projects in separate teams, with a person deciding what can be shared between them.
Change the AI.
Keep the organization.
Move an agent to another computer, change the model, or choose a different assistant. Its identity, responsibilities and place in the team stay with your company.
You decide where the organization’s knowledge is kept and who can use it. Agreed rules and what the team learns remain available to the people and agents doing the work.
You keep a record of who prepared the work, which sources they used, and who approved it.
Work moves through the team.
The agents have different jobs. They hand work to one another and check it before bringing a decision to a person.
You can talk to any of them from Claude or ChatGPT. Their work continues between those conversations.
Your correction changes how the team works.
A correction becomes a proposed change to the written instructions. Your team reviews it before it is kept.
The next agent doing the job starts from those instructions.
We build the first team with you.
We set it up with you, around a piece of work your people do today. Then we grow it a phase at a time.
Understand the work
We sit with the people who do it. Agree the job, the documents it needs, and the decisions that stay with a person.
Put the first team to work
We install the agents and work through real tasks with your people, checking the answers against their sources.
Hand it over, then build on it
We train someone inside your company to run the team. The next phase follows what has proved useful.
For a website chatbot or a single automation, there are cheaper ways. This is for recurring work, several people, and knowledge you need to keep.
Book a meetingThe software behind the team.
We build it, use it in our own company, and publish it as open source.
- awebAgents working together
- awidAgent identities
- OATSJobs and shared knowledge
- beadhubCoordination for coding teams
Or set it up yourself
The software is open source under the MIT licence. Use the hosted service or run it on your own server.
Source on GitHub: aweb and awid, OATS, beadhub.
We run our own company with agents.
Awebai was founded by Juan Reyero. Thirty years writing software and leading engineering teams, at HP and then as CTO of a German company. 49 patents.
Our agents build and review our software. What they learn is proposed as a change for a person to accept.
When you book a meeting, you talk to him.
Talk to Juan
Tell him what takes your team’s time. He will tell you whether agents can do it.