Automating Odoo CRM with AI: Lead Scoring, Notes and Follow-Ups

AI can take over the three chores that keep an Odoo CRM pipeline out of date: judging which leads deserve attention, writing up what happened, and remembering to follow up. An agent connected to Odoo reads each lead and its history, records a score with reasons, posts notes on the record and drafts the next message for the salesperson to send. The selling stays human. The admin around it does not have to.

Key takeaways

• The agent works on crm.lead records through Odoo's API, as its own user with CRM-only rights.
• Useful AI lead scoring gives a score and a written reason, based on criteria your sales team has agreed.
• Notes and call summaries go on the record's chatter, so the full history lives in Odoo and not in someone's inbox.
• Follow-ups are best drafted by AI and sent by a person, at least until you trust the output.

Why CRM data goes stale

A CRM is only useful if it reflects reality, and keeping it current is work nobody enjoys. Salespeople are paid to talk to customers, so logging calls, updating stages and scheduling the next activity get done late or not at all. Managers then forecast from a pipeline that is partly fiction, and leads that needed a second touch are forgotten.

Odoo's automation rules help with some of this, for example by creating an activity when a stage changes. What rules cannot do is read an enquiry or turn a rambling call recap into a clear summary. That reading is where an agent earns its place.

1. Lead capture and enrichment

Leads arrive through web forms, emails, chat and messaging apps, and often as a few unstructured lines. The agent's first job is to turn each one into a clean record.

• It extracts the name, company, contact details, the product or service of interest and any stated timeline.
• It searches res.partner to see whether this person or company already exists, and links the lead instead of creating a duplicate contact.
• It checks for an open opportunity for the same customer and flags it, so two salespeople do not chase the same deal.
• It fills the fields your team uses, such as source, tags and a short description, and leaves blank anything it could not find. It must not guess.

2. Lead scoring you can read

A score is only helpful if salespeople believe it. A bare number with no explanation gets ignored. The approach that works is a rubric the sales team writes, applied by the model, with the reasoning recorded.

Build the rubric first

Sit down with your best salespeople and list what separates a good lead from a poor one. Typical criteria:

• Fit. Is the company the type and size you serve? Is it in a region you cover?
• Need. Did they describe a specific problem, or only ask for a brochure?
• Timing. Is there a deadline or a stated start date?
• Authority. Is the contact a decision maker, or gathering information for one?
• Engagement. Did they reply, book a call, or go quiet?
• History. Are they an existing customer with past orders in Odoo?

How the agent applies it

For each new or updated lead, the agent reads the lead, the message thread and the customer's order history, then returns a structured result: a rating for each criterion, an overall priority and two or three sentences explaining why. A tool writes the priority to the lead and posts the explanation as an internal note. A salesperson opening the lead sees, for example, "High priority: existing customer, asked for a quotation for a named product, wants delivery this month. Contact is the purchasing manager."

Rules, statistics or a language model?

Method	Strength	Limit
Fixed rules (points per field)	Transparent and predictable	Cannot use the content of free-text messages
Statistical scoring from past deals	Learns from your real outcomes	Needs a large, clean history of won and lost deals
Language model with a rubric	Reads messages and explains its reasoning	Reflects your rubric, not measured probabilities; needs spot checks

These can be combined. An agent adds most where the signal is in the text.

One caution: a model's score is a judgement against your criteria, not a calculated chance of closing. Present it as a priority, and compare it against real outcomes every few months to see whether the rubric needs changing.

3. Notes that write themselves

Every lead in Odoo has a chatter, the running log of messages, notes and activities. An agent can keep it complete.

• Call and meeting summaries. Given a transcript or a salesperson's rough notes, the agent posts a structured summary: what was discussed, what the customer needs, objections raised, agreed next steps.
• Email thread digests. For a long back-and-forth, it posts a short "where things stand" note.
• Handover briefs. When a lead changes owner, it writes a brief so the new owner does not have to read forty messages.

Recording calls is subject to consent rules that vary by country, so settle that policy before transcribing anything.

4. Follow-ups that happen

Most pipelines leak at the follow-up stage. An agent closes that gap in two ways.

First, it makes sure every open opportunity has a next step. A scheduled job lists opportunities with no planned activity or with no contact for a set number of days. For each, the agent reads the history and schedules a suitable activity for the owner, such as "call to confirm the revised quotation was received".

Second, it drafts the message. Using the lead's history and your tone guidelines, it writes a short, specific email that refers to the last conversation instead of a generic "just checking in". The salesperson edits and sends it from Odoo, so the reply lands back on the same record.

Fully automatic sending is reasonable for a narrow set of messages, such as an acknowledgement of a new enquiry or a meeting reminder. For anything persuasive or commercial, keep a person in the loop. Our article on AI sales agents looks at the qualification conversation itself.

How the agent connects to Odoo CRM

The setup is small. The agent has its own Odoo user in the Sales and CRM groups only, and a short list of tools:

• Search and read leads and opportunities.
• Search contacts and read a customer's past orders.
• Create a lead.
• Update a limited set of fields: priority, tags, description.
• Post an internal note.
• Schedule an activity for a user.

It runs on triggers: a new lead is created, a message arrives on a lead, or a nightly schedule fires. Every action it takes appears in the chatter under its own user name, so the team can see exactly what the agent did and correct it. This is a typical first build for AI agents for Odoo and ERP because the risk is low and the benefit shows up in daily use.

What to keep with your salespeople

• Marking opportunities Won or Lost, and the lost reason.
• Expected revenue, probability and close dates used for forecasting.
• Prices, discounts and terms in any outgoing message.
• Reassigning leads between people, unless the rule is fixed and agreed.
• The decision to stop pursuing a lead.

Related guides

• What Is an AI CRM? Features, Use Cases and How to Choose
• HubSpot vs Zoho vs Odoo CRM: Which Works Best with AI?
• AI Meeting Notes to CRM: Automating Summaries, Tasks and Follow-ups
• Odoo Accounting and AI: Automating Vendor Bills and Invoices
• Service: AI Agents for Odoo and ERP

Frequently asked questions

Does Odoo already do lead scoring?

Odoo includes lead scoring and assignment capabilities in some editions and configurations. Check what your installation offers first. An AI agent complements it by reading free-text messages and writing the reasoning onto the record.

Will the agent send emails to customers on its own?

Only if you allow it. The usual starting point is drafts that a salesperson reviews and sends. Automatic sending can be enabled later for low-risk messages like acknowledgements.

How do we know the scores are any good?

Review a sample each week with the sales team and note where they disagree. Over time, compare priorities with what actually closed. If high-priority leads are not converting better than others, the rubric needs work.

Sources and further reading

Primary documentation and research behind this guide.

• Odoo 18 documentation: External API
• HubSpot: Understand the lead scoring tool

A CRM that updates itself gives managers a pipeline they can trust and gives salespeople their time back. If you want to see what this would look like on your own Odoo database, Growmerz can walk through it with you: book a free AI audit.