Sale Order Enhancements

The module extends sale.order with AI fields and actions:

• ai_score: close probability score from 0 to 100.
• ai_score_color: status color used by the score widget.
• ai_notes: generated quotation notes.
• ai_notes_tone: tone used to generate notes.
• ai_suggestions_json: raw AI response data for product suggestions.
• ai_suggestion_ids: structured product suggestion records.
• ai_insight: short AI insight for the current quotation.
• ai_recommendations: JSON list of sales recommendations.
• ai_last_analyzed: last AI analysis timestamp.
• customer_insight_id: link to customer intelligence.

Quotation Form UI

The quotation form receives:

• AI Analyze action button.
• AI score badge in the quotation header.
• AI Assistant notebook page.
• Floating AI assistant panel.
• Product suggestion list.
• AI notes editor.
• Customer insight action.

Settings UI

The module adds an AI Sales Assistant block in Odoo Settings with:

• Provider selection.
• OpenAI configuration.
• Google Gemini configuration.
• Custom model ID support.
• Test connection action.
• Auto-analysis toggles.
• Default note tone.
• Token usage counter.
• Monthly token limit.

Menus

The module adds an AI Sales menu with:

• Customer Insights.
• Product Suggestions.
• AI-analyzed quotations.
• High-risk deals.
• Hot deals.
• AI settings shortcut.

Dependencies

The module depends on:

• sale_management
• sale
• sale_crm
• product
• crm
• web

OpenAI

To use OpenAI:

1. Open Settings -> AI Sales Assistant.

2. Select OpenAI (ChatGPT).

3. Enter the OpenAI API key.

4. Select an OpenAI model.

5. Keep API Base URL as:

   https://api.openai.com/v1
   

6. Click Save.

7. Click Test Connection.

OpenAI-Compatible Gateways

For OpenAI-compatible gateways, local routers, proxies, or custom LLM endpoints:

1. Select OpenAI (ChatGPT).

2. Enter the gateway API key.

3. Set API Base URL to the gateway URL, for example:

   http://localhost:20128/v1
   

4. Select a listed model or choose Custom Model ID.

5. Enter the gateway model ID.

6. Save and test the connection.

The module calls:

POST {API Base URL}/chat/completions

Google Gemini

To use Google Gemini:

1. Open Settings -> AI Sales Assistant.
2. Select Google Gemini.
3. Enter the Gemini API key from Google AI Studio.
4. Select a Gemini text model.
5. Click Save.
6. Click Test Connection.

Recommended free-tier model:

Gemini 2.5 Flash Lite

The module calls:

POST https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent

Custom Gemini Model

If Google AI Studio exposes a model that is not yet listed:

1. Select Custom Model ID.
2. Enter the exact Gemini API model ID.
3. Save and test the connection.

Customer History Score

Based on confirmed or done sale orders from the same customer:

• 10 or more orders: high score.
• 5 or more orders: strong score.
• 2 or more orders: medium score.
• 1 order: low-medium score.
• No orders: new customer score.

Order Value Score

Compares the current quotation amount with the customer's historical average order value.

The score is higher when the quotation is close to the customer's usual order size and lower when the quotation is much higher than historical behavior.

Discount Score

Uses average discount across quotation lines.

The module treats moderate discounts as useful sales leverage, but very large discounts can lower confidence because they may indicate risk.

Response Time Score

If the quotation is linked to a CRM opportunity, the module compares:

• CRM lead creation date.
• Quotation creation date.

Fast quotation response improves the score.

Product Popularity Score

Checks how often the quotation products were sold in confirmed or done sale orders.

Popular products increase confidence. New or rarely sold products lower confidence.

Customer Engagement Score

Uses customer profile completeness and recent engagement:

• Email.
• Phone or mobile.
• Address.
• Company customer.
• Recent messages.

Adjustment Score

The module adds small business-rule adjustments:

• Customer has email and phone.
• Quotation has a valid expiration date.
• Quotation has order lines.
• Discount factor is favorable.
• Zero-amount quotations are penalized.

Risk Level

Final score maps to:

• high_chance: 70 and above.
• medium: 40 to 69.
• risky: below 40.

Context Sent to AI

The module sends:

• Customer name.
• Customer industry.
• Current quotation products.
• Last customer purchases.
• Similar customer purchases.
• Best-selling products.

Expected AI Response

The AI is instructed to return valid JSON:

{
  "suggestions": [
    {
      "product_name": "Product Name",
      "reason": "Why this product fits",
      "type": "upsell",
      "confidence": 85
    }
  ],
  "insight": "One sentence insight about this customer"
}

Suggestion Types

Supported suggestion types:

• upsell
• cross-sell
• bundle
• repeat

Validation

Before creating suggestion records, the module validates:

• Suggestion type.
• Confidence value.
• Product name length.
• Product match in product.product.

Adding Products

Users can click Add on a suggestion. If a matching product is found, the module creates a new sale.order.line with quantity 1 and marks the suggestion as added.

Chat Context

When a user sends a chat message, the module sends the AI:

• Customer name.
• Quotation amount.
• First quotation products.
• Current AI score.
• Quotation state.
• User message.

The assistant is instructed to give short, practical sales guidance.

Frontend Files

Main frontend files:

static/src/js/ai_panel/ai_sales_panel.js
static/src/js/ai_panel/ai_sales_panel.xml
static/src/js/ai_score_widget/ai_score_widget.js
static/src/js/ai_score_widget/ai_score_widget.xml
static/src/scss/ai_panel.scss

Buying Frequency

The module calculates average days between confirmed orders:

• Weekly.
• Bi-weekly.
• Monthly.
• Quarterly.
• Yearly.
• Irregular.

Price Sensitivity

The module calculates average discount requested:

• More than 15 percent: high price sensitivity.
• More than 5 percent: medium price sensitivity.
• Otherwise: low price sensitivity.

Preferred Products and Categories

The module counts products and categories from confirmed or done sale orders and stores the top results.

Best Contact Day

The module counts historical order days and picks the weekday with the highest frequency.

Security Groups

The module defines:

• AI Sales Assistant User
• AI Sales Assistant Manager

Access Rights

Users can:

• Read, create, and update AI suggestions.
• Read, create, and update customer insights.
• Use the regenerate notes wizard.

Managers can:

• Delete suggestions.
• Delete customer insights.
• Review all AI suggestions.

Record Rules

Sales users see suggestions linked to their own quotations. Managers can see all suggestion records.

API Key Storage

API keys are stored in ir.config_parameter and displayed as password fields in settings.

Do not publish screenshots that reveal API keys. If a key is exposed, delete it and create a new key.

OpenAI

OpenAI API billing is separate from ChatGPT Plus. A ChatGPT Plus subscription does not include OpenAI API credits.

For standard OpenAI usage:

API Base URL = https://api.openai.com/v1

For local routers or compatible gateways, use the gateway base URL.

Google Gemini

Gemini free tier quotas vary by model, project, region, and billing tier. A key can be valid while a model still fails because quota is 0, quota is exhausted, or the model is under temporary high demand.

If Gemini 2.5 Flash fails due to high demand, try:

Gemini 2.5 Flash Lite

OpenAI: Quota Exceeded

Cause:

• OpenAI API credits are not active.
• API billing has no remaining balance.
• Project has no usable quota.

Fix:

• Add credits in OpenAI Platform.
• Check project billing.
• Confirm the API key belongs to the project with billing.

Gemini: Quota Exceeded

Cause:

• Free tier quota is exhausted.
• Selected model has 0 quota.
• Google project needs billing for that model.

Fix:

• Check Rate Limit in Google AI Studio.
• Use a model with non-zero RPM/RPD.
• Use Gemini 2.5 Flash Lite for free-tier testing.
• Enable billing if production stability is needed.

Gemini: High Demand

Cause:

• Google temporarily throttles the selected model.

Fix:

• Wait a few minutes.
• Switch to a lighter model.
• Avoid repeatedly clicking Test Connection.

Custom Gateway Not Working

Check:

• Gateway is running.
• Base URL ends with /v1 if it follows OpenAI convention.
• Odoo server can reach the gateway URL.
• API key is correct.
• Model ID exists in the gateway.

AI Suggestions Empty

Possible causes:

• No product history.
• No best-seller data.
• AI returned invalid JSON.
• API quota or provider error.

Fix:

• Add confirmed sale orders for demo data.
• Use a reliable model.
• Check Odoo server logs.

Models

ai.sales.service
ai.sales.suggestion
ai.customer.insight
ai.regenerate.wizard
sale.order
res.partner
res.config.settings

Important Methods

sale.order.action_ai_analyze()
sale.order.action_ai_generate_notes()
sale.order.action_ai_get_suggestions()
sale.order.action_ai_predict_score()
sale.order.get_ai_panel_data()
sale.order.ai_chat_message()
ai.sales.service._call_ai_api()
ai.sales.service._call_openai_api()
ai.sales.service._call_gemini_api()
ai.sales.service.predict_deal_success()
ai.sales.service.get_product_suggestions()
ai.sales.service.generate_quotation_notes()
ai.sales.suggestion.create_from_ai_response()
ai.customer.insight._compute_insights()

System Parameters

sm_ai_sales.ai_provider
sm_ai_sales.openai_api_key
sm_ai_sales.ai_model
sm_ai_sales.custom_model
sm_ai_sales.api_base_url
sm_ai_sales.gemini_api_key
sm_ai_sales.gemini_model
sm_ai_sales.gemini_custom_model
sm_ai_sales.auto_analyze
sm_ai_sales.auto_suggestions
sm_ai_sales.default_tone
sm_ai_sales.total_tokens_used
sm_ai_sales.monthly_limit