Skip to Content

Zen AI Agent

by
Odoo

$ 99.00

v 18.0 Third Party
Availability
Odoo Online
Odoo.sh
On Premise
Odoo Apps Dependencies Discuss (mail)
Lines of code 8224
Technical Name zenai_agent
LicenseLGPL-3
You bought this module and need support? Click here!
Availability
Odoo Online
Odoo.sh
On Premise
Odoo Apps Dependencies Discuss (mail)
Lines of code 8224
Technical Name zenai_agent
LicenseLGPL-3
ZenAI Agent  •  Odoo 18  •  Community & Enterprise  •  v18.0.2.0.0 $399 • LGPL-3
AI Agents for Odoo 18

Your ERP.
Now it
thinks.

ZenAI embeds LLM-powered agents directly into your Odoo backend. They read your live data, answer questions, schedule automated reports, and surface insights your team used to spend hours finding.

⚡ Install Now — $399 View Docs
At a glance
3
LLM providers
10
agent templates
0
write risk by default
10
max tool rounds / msg
depends only on base + mail
Everything Included

One module. Eight production-ready capabilities.

Multi-LLM Support
OpenAI (GPT-4o, o1), Anthropic (Claude 4 family), and Google Gemini — swap providers without touching agents. API model list fetched live.
Live Read-Only Tool Calling
Three read-only ORM tools: search_records, read_record, count_records. Up to 10 agentic rounds per message.
Agent Scheduler
Cron-backed automated reports on daily, weekly, or monthly schedules. Results emailed to chosen recipients and saved as conversations.
Token Budget Control
Set a monthly token cap per agent. Warn at a configurable threshold (default 80%) or hard-block when the budget is exhausted. Resets on the 1st of each month.
Conversation Export
Export any conversation to PDF or CSV for sharing, record-keeping, or offline review. Available from the conversation toolbar.
10 Agent Templates
Ready-to-deploy role templates: Sales Analyst, Finance Controller, HR Manager, Inventory Manager, Purchase Manager, Project Lead, Customer Support, Executive Assistant, IT Admin, and General Assistant.
Smart Context Pinning
Pin any Odoo record (invoice, order, employee…) to a conversation. Pinned records are injected into the system prompt automatically on every message.
Multi-Turn Memory Summary
Long conversations are automatically compacted: oldest messages are trimmed first while the system prompt and latest user message are always preserved, keeping token costs low.
Role-Based Access Control
Restrict each agent to specific users or security groups. Per-agent model allow-lists validated on every tool call. User and Manager roles included out of the box.
Why ZenAI?

What changes when your ERP can answer back.

Task Without ZenAI With ZenAI
Check overdue invoices Navigate → Accounting → filter → export → read (8–12 min) Ask in chat → instant answer (5 sec)
Find top customers this month Ask analyst → wait for report → review (hours) Agent queries sale.order → ranked list (seconds)
Daily stock report for the team Scheduled report email (stale, fixed template) Scheduler runs agent daily → live answer emailed automatically
HR leave summary HR manager logs in → opens leave module → checks manually HR agent reads hr.leave in seconds from any screen
Project task status Open Project → filter by deadline → scroll "Which tasks are overdue?" → done
How It Works

A message becomes a live data query in milliseconds.

1
User types
A question in the Copilot chat inside Odoo backend.
 →
2
Context built
System prompt + pinned records + conversation history assembled.
 →
3
LLM decides
Model picks the right tool: search, read, or count — with exact args.
 →
4
Odoo queried
ORM executes read-only query. Sensitive fields stripped. Result sent back.
 →
✓
Answer delivered
LLM composes a human answer. Loops back if more data needed (up to 10 rounds).
// actual round trace — "Who joined HR this month?"
msg in → "Who joined HR this month?"
round 1 → search_records(hr.employee, ["create_date",">=","2026-05-01"], fields="name,department_id,job_title")
✓ 3 employees found  execution_time: 38ms • logged to zenai.agent.action
msg out → "3 employees joined this month: Alice (Sales), Bob (IT), Carol (Finance)"
Agent Scheduler

Set it once. Get reports forever.

Schedule any agent prompt to run automatically — daily, weekly, or monthly — and email the result to your team.

 
Daily
"Show products below reorder point"
→ emailed every morning at 08:00
 
Weekly
"Top 10 deals closed this week?"
→ sent to sales team every Monday
 
Monthly
"Revenue vs last month summary"
→ CFO inbox on the 1st of each month
Each scheduled run: ✓ saved as a conversation ✓ emailed to recipients ✓ last run status tracked ✓ run count logged
Use Cases

One module. Every department.

10 built-in templates get you started instantly — each scoped to only the models and users it needs.

Sales Analyst
Scoped to sale.order, crm.lead, res.partner
"Top 5 deals closing this week?"
"Show customers with no order in 90 days"
"What's the pipeline value by stage?"
Finance Controller
Scoped to account.move, account.payment
"Overdue invoices above $10,000?"
"Revenue this month vs last month?"
"Which customers owe the most?"
HR Manager
Scoped to hr.employee, hr.leave
"Who has pending leave requests?"
"Employees by department?"
"Who joined this month?"
Inventory Manager
Scoped to product.product, stock.picking
"Products below reorder point?"
"Pending delivery orders today?"
"Top 10 products by quantity?"
Purchase Manager
Scoped to purchase.order, res.partner
"Purchase orders pending delivery?"
"Top suppliers this quarter?"
"Orders not confirmed this week?"
+ 4 More Templates
Project Lead — project.task
Customer Support — res.partner
Executive Assistant — full access
IT Administrator — res.users
LLM Providers

Every provider. Pick the right model for each job.

Swap providers without touching your agents. Live model list fetched directly from the provider API. API key required.

 
AI
OpenAI
GPT series • models fetched live
Industry-standard tool calling. Supports GPT-4o, o1, and the full GPT-4 family via the official API. Model list is fetched live from the API.
Default Models
gpt-4o gpt-4-turbo o1 o3-mini
✓ Tool Calling ✓ Vision ✓ Live Model Fetch
 
C
Anthropic
Claude 4 family
Native Messages API with full tool_use blocks, consecutive-role merging, and multimodal image support handled automatically. Claude 4 models available.
Available Models
claude-opus-4 claude-sonnet-4 claude-haiku-4 claude-3.5-sonnet
✓ Tool Calling ✓ Vision ✓ Extended ctx
 
G
Google Gemini
Gemini 2.x family
Via Google’s OpenAI-compatible endpoint — zero extra code. Gemini 2.0 Flash delivers fast, cheap responses with full tool calling and 1M context.
Available Models
gemini-2.5-pro gemini-2.0-flash gemini-1.5-pro
✓ Tool Calling ✓ Vision ✓ 1M ctx
* Streaming not yet implemented — responses delivered on completion.
Security

Locked down.
By design.

Every tool call is read-only. Sensitive fields are stripped at the ORM layer before any data reaches the LLM. Each action is logged in a separate DB cursor — so audit records commit even if the main transaction fails.
✓ Read-only ORM tools — search, read, count only. No writes, ever.
✓ Sensitive field exclusion — password, api_key, token, secret auto-stripped before LLM sees data.
✓ Model allow-listing per agent — validated on every tool call. Or enable All Models for full read access.
✓ Conversation isolation — IR rules ensure users can only see their own conversations.
✓ Role-based groups — User + Manager roles. Per-agent user and group access lists.
✓ Full audit trail — tool name, args, status, error, execution time (ms). Committed in separate DB cursor.
Technical
Modulezenai_agent
Version18.0.2.0.0
Odoo Version18.0
LicenseLGPL-3
Price$99 USD
Python dependencyrequests
Odoo dependenciesbase, mail
Architecture
FrontendOWL (Odoo 18)
EditionsCommunity & Enterprise
DeploymentOnline • .sh • On-Premise
ORM toolssearch / read / count
Max tool rounds10 per message
Agent templates10 built-in
Scheduler typesdaily / weekly / monthly
FAQ

Questions we actually get asked.

Can agents write to Odoo?
No. The three tools — search_records, read_record, count_records — are all read-only. No create, write, or unlink calls are made under any circumstance.
Does it work on Community Edition?
Yes. ZenAI only requires base and mail. No Enterprise-only modules needed.
How does the Token Budget work?
Set a monthly limit per agent. ZenAI warns at a configurable threshold (default 80%) and can block requests when the budget is exhausted. The counter resets automatically on the 1st of each month.
What happens if the LLM hits the token limit?
Conversation history is trimmed oldest-first. The system prompt and the current user message are always preserved to maintain continuity.
Can I restrict which models an agent sees?
Yes. Each agent has an explicit allow-list of ir.model records, validated on every tool call. Use the All Models toggle for full read access.
Does Anthropic require special handling?
Yes — and we handle it. ZenAI converts OpenAI message format to Anthropic’s content blocks, merges consecutive same-role messages, and maps tool_use results back automatically.
Are tool actions logged even if the chat fails?
Yes. Each tool call is committed in a separate database cursor immediately — independent of the main transaction. Audit records survive even if the chat request rolls back.
Can I export conversations?
Yes. Any conversation can be exported to PDF or CSV directly from the conversation toolbar — useful for sharing reports or keeping records.

Your ERP already has the answers.

ZenAI teaches it how to talk.

⚡  Install ZenAI Agent — $99
ZenAI Agent  •  Odoo 18  •  v18.0.2.0.0  •  LGPL-3  •  $99 USD

Please log in to comment on this module

  • The author can leave a single reply to each comment.
  • This section is meant to ask simple questions or leave a rating. Every report of a problem experienced while using the module should be addressed to the author directly (refer to the following point).
  • If you want to start a discussion with the author or have a question related to your purchase, please use the support page.