TipsMay 8, 20269 min read

7 concrete AI use cases for Dolibarr in 2026

AI in an ERP is no longer a gimmick. Here are 7 cases that save hours every week, already deployed in production at our clients.

AI in an ERP: what really changes

For two years, AI was a buzzword bolted onto existing screens to look modern. In 2026 we have truly proven uses in production on hundreds of Dolibarr instances. The rule is simple: an AI use case is useful if it saves at least 30 minutes per day for a user, or if it improves the quality of a repeated decision. The seven cases below pass this rule. They are not demos, they are automations in place at our clients that have collectively freed hundreds of work hours per month.

Utility test: 30 min/day saved or better repeated decision
Cases from real deployments, not demos
All compatible with Dolibarr 16+ via DolicraftAI
Multi-provider choice: OpenAI, Anthropic, Mistral, local Ollama
Granular permissions: each case toggleable per role

Case 1 — Client emails drafted in 5 seconds

A rep needs to send a follow-up email after a client call. Instead of starting from scratch, they click the AI icon in the editor, type 'confirm the meeting date, refocus on tech criteria, ask who attends client side' and get a polished email ready to send. Five seconds vs three minutes. Multiplied by 15 emails a day, that's 45 minutes saved per rep daily. The module also offers rewrite actions: change tone (professional, friendly, concise), translate to six languages, fix grammar and suggest a subject.

Generation from short instructions
Rewrite with tone change
Translate FR/EN/ES/DE/IT/PT
Spelling and grammar fix
Automatic subject suggestion

Case 2 — Bulk SEO product descriptions

For e-commerce, hand-writing product descriptions is a bottleneck. AI generates 50 descriptions in 10 minutes by varying angle (commercial, technical, SEO-optimised on target keywords). E-commerce shops on Dolibarr typically save 2 days/month on this task. Quality is good enough for direct publication on 80% of products and only needs review for the 20% strategic ones.

3 angles: commercial, technical, SEO
Bulk generation via multi-product selection
Configurable target SEO keywords
Typical saving: 2 days/month for an e-commerce shop
Review only for strategic products

Case 3 — Sentiment analysis on support tickets

DolicraftAI automatically analyses each new support ticket and tags it with a sentiment (positive, neutral, negative, urgent) and a priority level. Negative or urgent tickets jump to the top of the support queue, defusing dissatisfaction before it escalates. On 200 tickets/month, this typically prevents 3 to 5 major escalations per month and mechanically improves the customer satisfaction score.

Automatic sentiment detection on ticket creation
Confidence score and priority level
Negative tickets sorted to the top
30-50% reduction in escalations
Compatible with any input channel (mail, form, API)

Case 4 — One-click client summary

Before a client meeting, the rep wants to know: who is this client, what did they buy, do they pay late, were there support incidents? Reading five tabs and 30 history lines takes 10 minutes. With the 'AI Summary' tab on the third-party card, AI produces a structured digest in 5 seconds: financial health, payment behaviour, relationship history, detected risks, actionable recommendations. Also great for a new rep picking up an existing portfolio.

Summary generated on demand on the third-party card
Sections: finance, payment, relationship, risks, recommendations
Ideal before meetings or portfolio handover
Typical saving: 8 minutes per meeting
Compatible with custom extrafields

Case 5 — SQL queries in natural language

The CEO asks: 'how many invoices unpaid over 30 days?'. Instead of asking a developer to write SQL, they type the question in the AI bar and get the answer formatted as table and chart. The module translates the question to read-only SQL, runs it in a secure sandbox, and returns the result. This democratises data access for non-technical users without risk to the database.

Plain-language business questions translated to SQL
Secure read-only execution (never writes)
Result formatted as table or chart
Democratises data access for non-technical users
Audit log of all executed queries

Cases 6 and 7 — Contract generation and contextual chat

Sixth case: contract generation. From a client card, AI produces a complete contract draft (services, schedule, clauses) you can edit before converting to a PDF auto-attached to the contract object in Dolibarr. Seventh case: the contextual AI chat available on every Dolibarr page. It knows what you are working on (invoice, quote, client) and adapts answers; you ask 'draft a thank-you note for this client' and the email appears with the right variables pre-filled. For organisations with strong confidentiality constraints, the module works in 100% local mode via Ollama, with zero data sent to an external cloud.