Odoo WhatsApp AI integration connects a WhatsApp Business API account to Odoo CRM, sales and support modules using AI intent classification to route messages, create leads, trigger workflow automations and send templated responses automatically. UK businesses using this integration reduce manual WhatsApp response time from 4 hours to under 2 minutes for routine enquiries.
Last updated: May 2026
Odoo WhatsApp AI integration is a bidirectional connection between your WhatsApp Business API account and your Odoo ERP, with an AI layer that classifies incoming messages, decides what action to take, and either responds automatically or creates the right Odoo record for human follow-up. It is not the same as Odoo's built-in WhatsApp module, which allows Odoo users to send template messages from within Odoo. The AI integration goes further: it listens to inbound messages, understands what the sender wants, and acts on that understanding inside Odoo without human intervention for routine message types.
The integration architecture has three layers. The first is the WhatsApp Business API connection: a webhook listener receives every inbound message from your WhatsApp Business number and passes it to the processing pipeline. The second is the AI intent classification layer: a large language model using OpenAI function calling reads the message, determines the sender's intent, new enquiry, price request, appointment booking, order status query, complaint, or general conversation, and extracts any structured data present in the message such as name, email, service interest, preferred time and order reference. The third is the Odoo integration layer: based on the classified intent, the system takes action in Odoo using the JSON-RPC API, creating a CRM lead, updating a contact record, triggering a sales workflow, querying an order, or creating a support ticket in Helpdesk.
The Odoo modules involved are Discuss (which provides the native Odoo WhatsApp thread interface if using Enterprise), CRM (for lead creation and pipeline management), Sales (for quotation triggering), Contacts (for contact creation and deduplication), and optionally Helpdesk (for complaint and support routing) and Calendar (for appointment booking). The integration is built on top of Odoo's official API, meaning it works alongside your existing Odoo setup without modifying core code and persists through Odoo version upgrades.
What messages are handled automatically vs. routed to a human:
The system is configured with a set of intent categories and associated automation rules during the implementation phase. Routine enquiries, requests for a quote, questions about services, requests to book a callback or appointment, order status queries, are handled with automatic responses and Odoo record creation. Complex or ambiguous messages, complaints, and any message the AI classifies with low confidence are routed to a human agent queue in the Odoo Discuss module or your preferred channel. The handover is smooth: the human agent sees the full conversation history, the AI-classified intent, and the Odoo records already created, rather than starting from scratch.
Template messages and the 24-hour session window:
WhatsApp Business API enforces a strict two-tier messaging model that every integration must respect. All outbound proactive messages, the kind your system sends first to a contact, must use Meta-approved message templates. These are submitted to Meta for review before use and typically approved within 1 to 5 business days for standard business content. Free-form messaging, where you can send any text without a pre-approved template, is only permitted within a 24-hour customer service window that opens when a contact messages you first. Once 24 hours pass from the customer's last inbound message, that window closes: the next message from your system must again be a Meta-approved template. This matters operationally because it means a contact who messages on Friday at 5pm and receives no human follow-up until Monday cannot receive a free-form reply: Monday's first message must be a template. Building the integration around this constraint, rather than discovering it after go-live, is essential for compliant operation.
AI intent classification is the core intelligence of the Odoo WhatsApp integration. When a message arrives from a WhatsApp contact, the LLM reads the full message text and any conversation history to provide context, then performs a structured classification task: it identifies the primary intent, extracts any structured data present in the message, and returns a JSON object with the classified intent, extracted fields, and a confidence score.
The classification uses OpenAI function calling, which forces the model to return a structured JSON response rather than free-form text. This is essential for automation: the downstream Odoo integration needs clean structured data, not prose. On standard UK business enquiry types, appointment requests, pricing queries and service availability questions, intent classification accuracy using GPT-4o function calling reaches 91 to 94%. The function definition specifies the intent categories your business uses, the fields to extract for each intent, and any validation rules, for example a UK mobile number must match the 07XX pattern.
An example of the classification flow for a typical UK SME:
A contact messages: "Hi, I run a small restaurant in Harrow and I am interested in getting a chatbot for taking bookings. Can you tell me how much it costs and how long it takes to set up?"
The LLM classification returns: intent = new_lead, extracted fields = {business_type: "restaurant", location: "Harrow", service_interest: "chatbot", specific_question: ["pricing", "timeline"]}, confidence = 0.97.
The Odoo integration then creates a new CRM lead with the contact's WhatsApp number, business type, location and service interest; tags the lead with the specific question topics; triggers an automated WhatsApp response with a pre-approved template acknowledging the enquiry, stating typical pricing and timeline, and asking for their name and the best time for a call; and notifies the assigned sales person via Odoo internal message that a new qualified lead has arrived with the conversation pre-populated. A new CRM lead created from a WhatsApp conversation is available in Odoo within 3 to 5 seconds of conversation completion via webhook, with no manual data entry involved.
The response time improvement from 4 hours to under 2 minutes for routine queries is not because a human responded faster. The AI sends a meaningful, personalised first response within seconds of the original message, and the human follow-up starts from a warm, already-qualified lead rather than a cold inbox message. For appointment requests and pricing queries, which together account for the majority of inbound WhatsApp messages at most UK SMEs, the AI handles the entire first exchange without involving a person at all.
Handling conversation context:
The AI maintains conversation context across a WhatsApp session. If a contact sends multiple messages in a conversation, the classification engine reads the full thread to understand the evolving intent. A contact who starts by asking about pricing and then mentions they have an urgent project this week would be classified differently from a contact who mentioned the same query with no urgency signal. The context window for the LLM is configured to include the last N messages, balancing accuracy against API cost.
Confidence thresholds and escalation:
Every classification returns a confidence score. Configurations typically set a threshold of 0.85: messages classified above this threshold proceed to automated handling; messages below are routed to a human agent queue with the AI's best-guess classification shown as context. This prevents the system from taking confident wrong actions on ambiguous messages, while still automating the clear-cut majority. In practice, 70 to 80% of inbound WhatsApp messages from a business context clear the 0.85 confidence threshold on day one; this rises to 85 to 90% after two to four weeks of calibration using the specific message patterns your contacts send.
The most valuable use cases for Odoo WhatsApp AI in UK businesses fall into three categories: inbound lead capture for service businesses, order and booking management for transactional businesses, and customer service deflection for businesses with high inbound query volumes. The integration delivers the highest ROI in businesses where WhatsApp is already an active customer channel but the volume of messages has outgrown the capacity of a person to handle them manually.
WhatsApp has 2.9 billion monthly active users globally. In the UK, 80% of adults aged 18 to 54 use WhatsApp as their primary messaging app, making it the dominant personal communications channel for the working-age population. For UK businesses serving this demographic, a missed or slow WhatsApp response carries a higher commercial cost than a missed email, because the expectation of rapid response is embedded in how people use the platform.
Inbound lead capture for service businesses:
For UK SMEs offering professional or trade services, consultancies, law firms, accountancies, IT services, tradespeople, healthcare clinics, WhatsApp is increasingly used by prospective clients to make first contact. These businesses often have a team too small to staff a dedicated WhatsApp inbox, so messages sit unanswered for hours. The AI integration creates a CRM lead in Odoo within seconds of the enquiry arriving, sends an immediate acknowledgement, and classifies the lead by service interest and urgency so the sales or enquiries team can prioritise follow-up. The response time improvement from 4 hours to under 2 minutes applies directly to this use case and is measurable in conversion rate improvement: leads contacted within 5 minutes are substantially more likely to convert than those contacted after 30 minutes.
Appointment booking for healthcare and professional services:
When connected to Odoo Calendar, the integration enables appointment booking via WhatsApp. A patient or client messages to book an appointment, the AI identifies available slots from Odoo Calendar and presents them in a WhatsApp interactive message, the contact selects a slot, and the Calendar event is created in Odoo automatically. The contact receives a confirmation message. For practices and clinics handling 20 to 50 booking requests per week via WhatsApp, this eliminates an estimated 15 to 25 minutes of admin per booking.
Order status queries for product businesses:
For UK businesses with an Odoo Sales or eCommerce setup, a significant proportion of inbound WhatsApp messages from existing customers are order status queries: where is my order, when will my delivery arrive, can I change the delivery address. The AI integration queries the Odoo sale.order and stock.picking APIs in real time and returns the current order status, tracking information and expected delivery date in the WhatsApp reply, without any human intervention. Customer service agents are freed from handling repetitive status queries to focus on actual problems requiring human judgement.
Post-sale follow-up and upsell sequences:
Using outbound template messages, sent only to opted-in contacts in compliance with PECR, the integration triggers Odoo-driven WhatsApp sequences: a satisfaction check after a service delivery, a review request at the appropriate interval, a renewal reminder before a subscription expires, or a targeted upsell offer based on the customer's purchase history in Odoo. These sequences are triggered by Odoo automation rules: when a sale order reaches the done state and the product category matches a specific service type, the WhatsApp follow-up sequence starts automatically.
Parallel GoHighLevel integration:
For businesses running both Odoo and GoHighLevel, the same WhatsApp webhook delivers lead data to both platforms simultaneously. When a new lead intent is classified, the integration creates a CRM lead in Odoo and sends the same contact data to GoHighLevel via webhook, triggering GHL automation sequences in parallel. This is particularly useful where the Odoo CRM handles the delivery team's work and GHL handles the marketing automation side.
Odoo WhatsApp AI integration costs from £4,000 for a focused CRM and lead routing implementation with intent classification. Full multi-flow implementations covering lead capture, booking, order queries and outbound sequence triggering cost £6,000 to £12,000. The cost is driven by the number of intent categories and Odoo integrations in scope, the complexity of the response templates, and the number of workflow automations to configure in Odoo.
| Implementation Scope | Typical Cost Range | Timeline |
|---|---|---|
| CRM lead creation + intent routing + auto-response (2 - 3 intents) | £4,000 - £5,500 | 5 - 6 weeks |
| CRM + Calendar booking + order status queries (4 - 6 intents) | £5,500 - £8,000 | 6 - 7 weeks |
| Full multi-flow: CRM + Booking + Order + Helpdesk + GHL parallel (6+ intents) | £8,000 - £12,000 | 7 - 9 weeks |
| Enterprise: multiple WhatsApp numbers + team routing + CRM pipeline rules | £10,000 - £18,000 | 8 - 12 weeks |
Timelines include the Meta WhatsApp Business API approval process, which adds 1 to 2 weeks compared to a standalone AI chatbot implementation. Meta business verification takes 1 to 5 business days; phone number registration takes 24 to 48 hours after approval. If you already have an approved WhatsApp Business API account, the timeline reduces by one to two weeks.
Ongoing costs:
WhatsApp Business API message costs depend on message category and country. For the UK, conversation costs in 2026 are approximately $0.04 to $0.08 per 24-hour conversation window. LLM API costs for intent classification are approximately £0.001 to £0.003 per message classified. For a business handling 500 inbound WhatsApp conversations per month, total variable costs are under £60 per month. There is no ongoing platform licence fee for the integration code: you own the implementation.
ROI for a service business handling 100 WhatsApp leads per month:
At 100 WhatsApp leads per month with a 30-minute average handling time per lead, manual processing costs 50 hours per month. At a staff cost of £25 per hour, that is £1,250 per month. The AI integration reduces this to approximately 10 hours per month, handling only escalated and complex leads. Monthly saving: approximately £1,000. A £4,000 to £5,500 implementation recovers build cost in four to six months, after which the saving is recurring. Lead conversion improvements from faster response time are not included in this calculation and typically dwarf the staff cost saving in commercial impact.
UK GDPR and PECR compliance for Odoo WhatsApp AI integration requires careful design from the outset. The obligations differ significantly depending on whether a message is inbound (contact-initiated) or outbound (business-initiated), and the distinction between marketing messages and service messages is the most practically important compliance boundary to understand.
Inbound messages - lawful basis:
When a contact messages your WhatsApp Business number, the lawful basis for processing that communication and the data it contains is legitimate interests under UK GDPR Article 6(1)(f) for business enquiries, or contract performance under Article 6(1)(b) for existing customers making service or order queries. The processing is expected: the contact chose to message your business number. The data collected, message content, WhatsApp number, any personal information voluntarily shared in the message, should be stored in Odoo only to the extent necessary for responding to the contact and managing the resulting business relationship.
PECR: marketing messages vs. service messages - a concrete distinction:
The Privacy and Electronic Communications Regulations draw a hard line between two types of outbound WhatsApp message, and the compliance requirement is completely different on each side of that line. Service messages, order confirmations, appointment reminders, delivery notifications, responses to a specific query a customer raised, do not require prior explicit consent under PECR. They can be sent to any contact you have a legitimate service relationship with, provided you send them only as Meta-approved templates outside the 24-hour free-messaging window. Marketing messages, promotional offers, upsell sequences, product announcements, require explicit prior opt-in consent under PECR even for existing customers. WhatsApp is classified as a personal communications channel under PECR, which means the soft opt-in exemption that applies to email marketing (allowing you to email existing customers about similar products without fresh consent) does not apply. A UK business sending WhatsApp promotional messages to its customer list without explicit consent for each recipient is in breach of PECR, regardless of whether those customers are B2B or B2C. The integration is designed with this distinction built into the outbound automation rules: service message templates are grouped separately from marketing templates, and the marketing group requires a consent flag on the Odoo contact record before any message is triggered.
Data residency and Meta data processing:
WhatsApp processes message data on Meta's infrastructure, which includes servers in the United States and the European Union. Under UK GDPR, sending personal data to Meta's US infrastructure is a restricted transfer. Meta provides standard contractual clauses and a Data Processing Agreement for WhatsApp Business API users. You should review Meta's DPA, ensure SCCs are in place for UK-to-US transfers, and document this transfer in your Records of Processing Activities. The message content itself is end-to-end encrypted in transit; Meta's servers process metadata, sender number, timestamp, message type, rather than the content of encrypted messages.
Data retention in Odoo:
Message content and lead data created from WhatsApp conversations should be subject to your standard CRM data retention policy. For UK businesses, a reasonable retention period for CRM records is 3 to 5 years from last contact, with a review process for inactive contacts. Contacts who request erasure under UK GDPR Article 17 must have their WhatsApp-sourced CRM record deleted alongside any other personal data held in Odoo. The integration includes a documented process for handling erasure requests covering the WhatsApp conversation logs, the Odoo CRM record, and any LLM API cache.
Privacy notice:
Your privacy notice should describe that WhatsApp is used as a communication channel, that messages may be processed using AI classification tools, and that message data is stored in your CRM. A brief statement on your website WhatsApp widget or business profile is sufficient for inbound enquiries. For outbound marketing sequences, the consent mechanism itself serves as the disclosure: the opt-in language should clearly describe what messages the contact will receive.
Odoo WhatsApp AI integration takes 5 to 7 weeks from project start to go-live for a standard CRM and intent routing implementation. The timeline is longer than a comparable integration without WhatsApp (which would take 3 to 4 weeks) because of the Meta WhatsApp Business API approval process, which is outside our control and adds 1 to 2 weeks.
Week 1 - Discovery and Meta business verification initiation: We run a discovery workshop to define the intent categories, map the Odoo workflow actions, design the response templates, and confirm the PECR compliance approach for any outbound messages. Simultaneously, we begin the Meta business verification process. Meta requires a verified Facebook Business account, your UK Companies House registration number, a business address matching Companies House records, and a verified business phone number. Businesses with their Companies House details ready typically complete verification within 1 to 5 business days; phone number registration completes within 24 to 48 hours of business approval. We guide you through the submission to minimise delays.
Weeks 2 to 3 - Development: While Meta verification is processing, we build the webhook integration, intent classification pipeline, Odoo API integration layer, and response template system. We configure the Odoo CRM lead creation logic, contact deduplication rules, and any workflow automation triggers that fire based on classified intents.
Week 3 to 4 - Internal testing and template submission: We test the full pipeline with a test WhatsApp Business number. Simultaneously, we submit your approved response templates to Meta for review. Meta template approval takes 1 to 5 business days for standard business templates; templates with promotional content take longer. We advise on template wording to maximise the likelihood of first-submission approval.
Week 5 to 6 - Supervised go-live and calibration: Once Meta verification and template approvals are complete, we switch the integration to your live WhatsApp Business number. Your team monitors inbound classification results for the first two weeks, flagging any misclassifications. We refine the intent classification prompts and confidence thresholds based on the patterns in your real message data.
Week 6 to 7 - Handover and documentation: We produce operational documentation covering how to add new intent categories, how to submit new template messages, how to handle the exception queue in Odoo Discuss, and how to extract WhatsApp lead data from Odoo CRM for reporting. We provide a three-month support SLA for integration issues and any changes arising from Meta API updates.
The most common cause of timeline extension is delayed Meta business verification, particularly for businesses whose Facebook Business account name differs from the Companies House business name. We recommend starting the Meta verification process as the first step in the project to prevent it becoming a critical path blocker.
You need a WhatsApp Business API account. The WhatsApp Business app (available free from the App Store or Google Play) does not support webhooks, automated responses or third-party integrations. The Business API is a separate product accessed through a Business Solution Provider or directly via Meta. It requires Meta business verification and a phone number not already registered to the Business app. Some BSPs include a monthly fee of £30 to £80; direct API access via Meta is free aside from per-conversation message costs. We guide you through BSP selection or direct API setup as part of the implementation.
The AI handles text messages, which covers the majority of inbound WhatsApp business enquiries. It can also process voice messages (using OpenAI Whisper to transcribe audio to text before classification), images with text descriptions, and document attachments such as PDF quotes or invoices a contact sends for review. Interactive message responses, where you send a contact a list of options and they tap a selection, are handled natively by the integration and do not require AI classification. The system does not handle video messages in the initial implementation scope.
Yes. When a contact's message is classified as a pricing or quotation request and they provide sufficient information, product or service type, quantity, delivery requirements, the integration creates a draft Odoo sales quotation pre-populated with the extracted information and assigned to the relevant sales person for review and sending. The draft quotation appears in the Odoo Sales module in the existing customer's or new lead's record. This works best for businesses with relatively standard products or service packages where the key quotation variables can be extracted from a WhatsApp message reliably.
Meta business verification confirms that your WhatsApp Business API account is associated with a legitimate registered business. You provide your legal business name, which must match official registration, your business registration number, website URL, and a business phone or email for verification. Meta cross-references this against public business records. For UK businesses, Companies House registration details are the most reliable data source. Business verification typically takes 1 to 5 business days; phone number registration takes a further 24 to 48 hours after approval. Businesses where the Facebook Business account name differs from the Companies House name sometimes experience longer review times. Start the process early.
Messages where the AI classification returns a confidence score below the configured threshold (typically 0.85) are routed to the human agent queue in Odoo Discuss with a notification to the assigned team member. The agent sees the message, the AI's best-guess classification and a brief explanation of why confidence was low: ambiguous intent, unusual phrasing, or mixed intents in a single message. The agent handles the message manually and their response is logged for calibration. Over time, repeated message patterns that initially scored low confidence are added to the intent training data, improving the system's handling of your specific contact base.
Yes, using pre-approved Meta template messages. Outbound messages triggered from Odoo, appointment reminders, order confirmations, follow-up sequences, are sent using Meta-approved templates, which can include personalisation variables such as the contact's name, appointment time or order number drawn from Odoo records. The trigger is an Odoo automation rule: when a Calendar event status changes to confirmed, send a WhatsApp confirmation template to the attendee's contact number. Outbound session messages (free-form, within a 24-hour window) can also be sent by team members directly from the Odoo Discuss interface on Enterprise.
Contact information and lead data extracted from WhatsApp conversations is stored in Odoo CRM and Contacts, subject to your existing Odoo data retention and access control policies. Conversation history is stored in Odoo Discuss (if using Enterprise) or in your webhook log database. Raw message logs should be retained only as long as operationally necessary, typically 12 to 24 months for CRM conversation logs, and deleted per your retention policy. Meta's WhatsApp Business API DPA covers the transfer and processing of message metadata on Meta's infrastructure. You must ensure SCCs or the UK IDTA are in place for any UK-to-US data transfer via the API, and document this transfer in your ROPA.
Yes. If you are on Odoo Enterprise and using the native Odoo Discuss WhatsApp feature, this AI integration complements rather than replaces it. The AI integration adds inbound processing intelligence, reading and classifying incoming messages, creating CRM records, triggering automations, which the native Discuss WhatsApp feature does not provide. The integrations coexist at the API level: both use the same WhatsApp Business API account, with the webhook router directing inbound messages to the AI pipeline and the Odoo Discuss module handling outbound messages initiated by your team.
Odoo WhatsApp AI integration addresses one of the most persistent friction points in UK SME sales and customer service: the gap between customers who expect instant responses on WhatsApp and businesses that cannot staff a real-time WhatsApp inbox. The 4-hour-to-2-minute response time improvement is the headline figure, but the deeper value is in lead quality: every classified WhatsApp enquiry arrives in Odoo CRM with context, intent data and conversation history, so your team can follow up faster and more relevantly than if they were responding to a raw message. The compliance framework for UK GDPR and PECR is manageable with proper consent design for outbound marketing messages and a DPA in place with Meta; the line between service messages (no consent required) and marketing messages (explicit consent required) is the most important design decision in any UK WhatsApp integration. Five to seven weeks from project start to go-live, including the Meta verification overhead that most businesses underestimate. See our Odoo AI integration London page to start scoping your implementation, or our Odoo ERP implementation London page if you are also considering a broader Odoo rollout.
Softomate delivers Odoo AI integration for London businesses from £3,000. Book a free discovery workshop to scope your use case.
Deen Dayal Yadav is the founder of Softomate Solutions, a London AI automation agency based in Stanmore. He has delivered over 200 AI automation and Odoo integration projects for UK businesses.
More from the blog
Let us help
Talk to our London-based team about how we can build the AI software, automation, or bespoke development tailored to your needs.
Softomate Solutions Limited
Deen Dayal Yadav
Online
