Creating an agent
The four creation paths, choosing the AI model, writing a good mission, the trigger, the limits, and the dry run right after creation.
12 min readUpdated October 1, 2026
Four paths lead to the same result. Pick according to how much you already know about your need.
The four paths
| Path | When to pick it | Time |
|---|---|---|
| Start from a template | Your need resembles a common case | 2 minutes |
| Describe it in plain language | You know what you want, not how to set it up | 5 minutes |
| Configure it by hand | You want to control every rule | 15 minutes |
| Import a file | You are bringing over an agent exported from another workspace | 1 minute |
Start from a template
The Template gallery offers eight proven agents, with permissions already trimmed to the strict minimum. You pick one, adjust the mission and the schedule, and it is ready.
This is the recommended path for a first agent, even when the template does not match exactly: adapting a well-framed agent is easier than setting everything up yourself.
Describe it in plain language
With Describe in plain language, you write what the agent must do, and Aidoo proposes a full configuration: rewritten mission, trigger, rules, limits. If something is still unclear, it first asks you a few questions.
The description can be long, up to 40,000 characters: a full specification fits without splitting, and Aidoo condenses it into a mission of at most 20,000 characters. The proposed model is always Auto; you can pick a specific one later, from the agent's configuration.
Every Monday morning, find quotations above €10,000 sent more than two weeks ago with no answer, and schedule a follow-up activity for the salesperson in charge.
The proposal always goes through your review
Aidoo never creates the agent from the description alone. You see the proposed configuration, you adjust it step by step, then you create it. This generation consumes a few credits, like any model call.
Configure it by hand
The Blank form, step by step: Identity, Role & AI model, Trigger, Permissions, Limits & safety, Notifications & access, Review. Useful when your case is specific, or when you want to understand exactly what each setting does. The other paths go through the same steps, prefilled.
Import a file
An agent exports to the aidoo-agent/v1 format and imports elsewhere with Import a JSON file (aidoo-agent/v1). Handy for reusing a proven agent in another workspace. The export carries no Odoo identity and no secret: you designate the member again on import. The format, a full example and the reference of every field are in Importing and exporting an agent.
Writing a good mission
This is the setting that matters most. A vague mission produces an unpredictable agent.
Say precisely what to look for. "Overdue invoices" leaves too much room; "validated customer invoices, unpaid, past their due date" leaves none.
Say what to do with each item. The agent needs to know whether it adds a note, schedules an activity, sends an email, or simply counts.
Say how to conclude. "Finish with a summary: number of overdue invoices, total amount, top five customers concerned." Without that sentence, the reporting varies from one run to the next.
Bound the volume. "Process at most the 20 oldest" prevents a run from taking on 400 records the day your database grows.
The mission accepts up to 20,000 characters: enough for a detailed specification, with its rules, its edge cases and the expected output format.
The read-aloud test
Have someone unfamiliar with the need read the mission. If they ask you a question, the agent will hit the same question, and it will answer it on its own, its own way.
Choosing the AI model
By default, the model is Auto: Aidoo picks an economical model suited to agents and keeps it up to date, with nothing for you to set. It is the right choice for most missions.
If your mission requires it, pick a specific model. Each one shows three markers: Cost, Speed, Power. Cost translates directly into credits consumed.
| Type of mission | Suggested model |
|---|---|
| Listing, counting, summarising a list | An economical model |
| Writing a message adapted to context | A mid-tier model |
| Analysing, comparing, choosing between options | The most capable model |
Start with Auto or an economical model. If results are rough, move up one notch: that beats paying for a powerful model to count invoices.
During the free trial
Only economical models are available, and the trial is capped at 500 credits or 30 days, whichever comes first.
The trigger
Three modes: Scheduled, Event-driven or Conversational.
Scheduled: a ready-made rhythm ("Every weekday morning (8am)", "Every hour"…) or a Custom one: frequency, days, time. The custom rhythm accepts up to 5 slots for different times on different days, for instance Monday to Thursday at 3 p.m. and Friday at 1 p.m. (Add a slot). No tier bounds that frequency: the agent follows exactly the schedule you give it, in its own timezone, with at least 5 minutes between two runs. Operating hours optionally restrict automatic runs to certain days and time ranges.
Event-driven: you pick an Odoo model, the event type (creation or update), and optional conditions. Aidoo watches that model at a rhythm set by your tier: one hour on Starter, a quarter of an hour on Pro, five minutes on Business. That is the agent's maximum reaction delay.
For instance: model crm.lead, on creation, with no condition, to qualify every new lead. Or model sale.order, on update, with the condition state = sale, to react to confirmed orders.
Event-driven, at a date: with At a record date, the agent runs at the exact moment a date field of the record falls, brought forward or pushed back by the number of minutes you choose (at most 3 firing points, at most one week of offset). For instance: model calendar.event, 30 minutes before start to prepare a briefing.
Conversational: no trigger. The agent answers when someone talks to it, from Aidoo or from the Odoo bubble.
The Odoo identity
In the Identity step, the Agent's Odoo account field designates the account whose rights the agent will borrow. Only owners and admins can make this choice. Only accounts whose member has saved their personal Odoo token in Aidoo can be selected: the others are greyed out.
No extra Odoo licence needed
An agent runs on any existing Odoo licence: it borrows the account of a member already in place, with no extra user to create and no licence to pay for. That account sets its scope, so pick it according to what the agent must be able to touch.
Permissions: learning by default
You do not have to fill in the rules matrix at creation. The Learning mode box is ticked from the start: for 48 hours, the agent discovers on its own what it needs.
- It reads everything its Odoo account can see.
- A write on a model no rule covers is put on hold: you approve or reject it, from Aidoo or from the Odoo bubble. Nothing is written without you.
- Every access actually made is recorded in the observed accesses list, shown on the agent's page.
- After 48 hours, permissions are frozen: observed accesses become ordinary rules you can adjust. You can also freeze earlier, in one click.
Rules coming from a gallery template, a prompt-based generation or an imported file stay in place as starting rules: their writes run without approval, learning fills in the rest. Once permissions are frozen, an owner or admin can Restart learning from the agent's page for a new 48-hour window.
Untick the box to get the classic rules editor back, described in Rules and permissions.
Put it to work during the window
An agent that observed nothing ends up with no permission at all when frozen. Run it, chat with it, let it handle a few real cases: what it does during those 48 hours is what it will be allowed to do afterwards.
The same step offers optional capabilities, all off by default:
- Web access: search the web and read public pages.
- Human approval for writes: every creation, update or action in Odoo waits for your approval.
- Suggest emails (approved sending): the agent drafts Odoo e-mails, you review them before they are sent. Automatic sending is enabled after creation, from the agent's page.
- Document reading: read PDFs and images attached to records, scans included.
- Can ask a human a question: in autonomous runs, the agent stops on a question and resumes with your answer.
- Aidoo Drive: read and write files in the workspace's file space.
External apps (Slack, Gmail…) are covered in Integrations.
Notifications and access
The step before last sets where the agent reports: e-mail, Odoo chatter, Discuss channel. It also designates, right at creation, the Odoo users who will see the agent in the Aidoo bubble of their Odoo: your own account is pre-ticked if it is linked, and Select all ticks the whole team. A conversational agent can only be used in Odoo by the users ticked here. The list can be changed later from the agent's page.
The safety limits
The Limits & safety step gathers the guardrails, adjustable per agent:
| Limit | Purpose | Default |
|---|---|---|
| Max actions / run | Stops a loop from consuming endlessly | Unlimited (empty field) |
| Max credits / run | Caps the cost of a single run | 100 credits, also the minimum |
| Max duration (seconds) | Interrupts a run that drags on | 300 s, up to 4,500 s |
| Max writes / run | Blocks the agent beyond N writes | Unlimited (empty field) |
A write counts one call, not one record: updating 50 tickets in a single grouped call counts as 1. The write cap is still the most useful early on: set to 5, it lets you see what the agent changes before you give it free rein.
Automatic pause stops the agent after several failures in a row (3 by default). You can also set an Agent monthly budget (credits), on top of the workspace pool, and, if the agent sends its own e-mails, a Max automatic emails / run (10 by default).
Creating the agent
The Review ends with a single button, Create the agent: the agent is created and activated immediately. It receives its own API key, whose permissions derive exactly from its rules: no more, no less. During the learning window that key carries every operation, the rules engine deciding call by call; it is tightened when frozen.
If activation is refused (no Agents subscription, insufficient tier, active agent limit reached), the agent stays in Draft: its page shows the reason and the Activate button.
Testing safely
Do a dry run right after creation
No test is launched automatically. From the agent's page, Test (dry run) runs the mission for real, but simulates writes: the agent shows you what it would have created or changed, without touching Odoo. It is the only way to see how it interprets your mission before it has any effect. If you want to test before its first slot, or rework its mission first, pause it (Pause) while you try it out.
Every run drops its report into the agent's discussion thread: what it observed, what it actually wrote, and the figures that sum up its pass. That is the first thing to read after a test.
For the detail, the debug icon on a run bubble opens the full trace: every tool call, with its parameters, result, status and duration. If the agent reads the same record ten times or wanders onto an unexpected model, the mission needs rewording.
A run heading the wrong way, real or simulated, is interrupted with Stop: the tool in progress finishes, then the agent stops.
The first real run deserves a careful read. After that, the daily digest is enough.
After creation
Everything is then managed from the agent's page:
- Rename: click its name.
- Pause, Resume, Run now on demand.
- Duplicate: a draft copy, with the same configuration and its knowledge sources.
- Archive: the agent is deactivated and its API key revoked. Find it again with Show archived agents in the list: Unarchive puts it back in pause with its key reactivated.
- Quick commands and Collaborating agents are configured on the agent's page only, not in the creation wizard. See What is an agent?.
Next step
- Rules and permissions: bounding what the agent may touch
- Credits and consumption: measuring and controlling cost
- Best practices and troubleshooting: when an agent misbehaves