Let the platform do the work
SugarClubContact+1 (877) 842-7276

sales-i Release Notes

Overview

This page outlines the feature enhancements and fixed issues included in the latest release of sales-i. For past release notes, please refer to the sales-i Archived Release Notes page.

sales-i 25.2.5 (April 20, 2026)

Feature Enhancements

  • SugarCRM is now SugarAI: You will now see our new company name across the sales-i experience, reflecting our SugarAI brand refresh. Learn more about the SugarAI brand evolution in this SugarClub post.
  • Re-engineered AI-generated opportunities: We have rebuilt our AI and ML models from the ground up to deliver a new generation of AI-generated opportunities designed to be more relevant, actionable, and easier to understand. These improvements help teams identify the right opportunities faster and take more confident action based on data-driven insights.
    • AI-generated opportunities renamed to Recommendations: AI-generated opportunities are now called Recommendations, reflecting a clearer, more intuitive experience across sales-i.
    • Expanded eligibility: All instances with 12 months of sales data and an average of at least two different products purchased per customer are eligible for recommendations. The number of recommendations generated will still depend on your specific sales data.
    • More data analyzed over a longer period: Recommendations are now more powerful and relevant than ever. By leveraging a full year of product association data across multiple invoices, combined with recent customer sales activity from the past 30 days, we deliver smarter, more accurate recommendations that drive real value.
    • Suggested quantity: Recommendations now use your sales data to suggest how many of the recommended product to sell, helping you better understand the true estimated value of each recommendation.
    • Enhanced explainability: Recommendations now include clearer descriptions to explain why they were made and how to act on them.
    • Automatic expiration of Recommendations: Recommendations now expire after 30 days from creation to ensure only relevant recommendations are surfaced.
    • Smart product filtering: Certain products, such as delivery charges, are automatically excluded from recommendations.
    • De-duplication of AI-generated Recommendations: Any updates to existing AI-generated recommendations, such as changes to quantity or estimated value, are applied to the original recommendation rather than creating a new one, ensuring each recommendation remains a single, up-to-date record.
    • De-duplication of manually created recommendations: Manually created recommendations surfaced in SugarAI from sales-i alerts are now de-duplicated based on alert report changes, ensuring only valid results are shown.
    • Improved Recommendations user experience: Each recommendation is now displayed individually rather than being grouped by customer, making it easier to review and act on specific recommendations.
    • Clearer Associations settings: The Associated Products administration interface has been improved to make it easier to understand product associations.
    • Deprecation of Communities settings: The Communities administration settings are no longer used by the new AI and ML models and will no longer be available.

Fixed Issues

  • Alert reports using both "buying" and "not buying" product data conditions could fail to generate results for some customers. Reports using these conditions now generate results correctly.
  • Changing the display language on a customer record prevented reports on the record from loading. Reports on customer records now load correctly after changing the display language.
  • The "Last 6 complete months, but not yet bought this month" Customer Insight report incorrectly showed zero YTD revenue for customers with revenue. It now displays the correct revenue values for customers.
  • When creating an alert report, the "Customers in Report" value did not update and always displayed zero. It now updates correctly as the report is configured.
  • The Plot to Map feature displayed incorrect locations for some addresses. It now shows the correct locations for all valid addresses.

FAQs

Why does this change matter?

We’ve reengineered our AI and ML models to deliver smarter, more actionable recommendations. This update prioritizes quality over quantity so while you may see fewer recommendations, each one is better aligned with your sales data, more time-aware, and more likely to drive new opportunities for your business.

Are the new AI-generated recommendations available to all customers?

The new AI-generated recommendations will be available to all sales-i 25 customers that meet the eligibility criteria: 12 months of sales data with an average of at least two different products purchased per customer. The number of recommendations generated will still depend on your specific sales data.

What data is analyzed by the models?

The last 365 days of sales data are analyzed and aggregated monthly to identify product association patterns where customers frequently purchase products together.

The last 30 days of customer sales are then reviewed. If a customer has purchased one product in an identified association but not the related product, a recommendation is created.

How often does the recommendation process run?

The recommendation process runs each time we ingest your ERP data, which is typically once a day.

Is eligibility criteria checked once or continuously?

Eligibility is checked continuously. Each time the recommendation process runs, your data is reassessed.

Why are recommendations not being generated?

Recommendations will automatically be enabled for any sales-i 25 customer that meets the eligibility criteria: 12 months of sales data with an average of at least two different products purchased per customer. The number of recommendations generated will still depend on your specific sales data.

Here are a few common situations that may prevent recommendations from being generated:

  • Bespoke or custom products make up the majority of sales

  • No invoices, or very few invoices, in the past month

  • Product associations exist, but the products have already been sold to all relevant customers

Why am I seeing fewer or no recommendations after the release?

The new models run on default parameters, whereas the previous models may have used adjusted parameters. Additionally, the new models prevent duplicate recommendations, which may result in fewer records.

What happens to my existing AI-generated opportunities?

AI-generated opportunities created from our previous models will be replaced with AI-generated recommendations created from the new models. Activities and interactions linked to previous AI-generated opportunities will be preserved.

These recommendations deliver a new generation of potential opportunities designed to be more relevant, actionable, and easier to understand. These improvements help teams identify the right opportunities faster and take more confident action based on data-driven insights.

Will this affect historical reporting or dashboards?

Historical reporting and dashboards (excluding recommendations) will remain unchanged.

Do I need to take any action after the release?

No action is required.

Do I need to act on every recommendation?

No. You can dismiss a recommendation if you feel it is invalid.

If the product associated with the recommendation is sold to the related customer before the expiry date, the recommendation is marked as “Won.” “Won” recommendations appear only in the “Won” tab on the Opportunities card of the customer record.

If the recommendation reaches its expiry date before the associated product is sold, it is deleted. The default expiry window is 30 days.

What happens when a recommendation expires?

When a recommendation expires, it is deleted. If the recommendation is still valid the next time the process is run, it will be recreated.

How are the suggested quantity and estimated value of a recommendation calculated?

The suggested quantity is based on our models analyzing your customers' buying patterns.

The estimated value is the average unit price multiplied by the suggested quantity.

Can I control which products are used to create product associations?

Yes, you can manually exclude products via your Associations administration settings in sales-i.

Can I adjust the default parameters used by the models?

Yes, you can reach out to support to adjust any of the following model parameters:

  • How confident the model needs to be to create a product association

  • How many days are analyzed in aggregate when analyzing the last 365 days of sales data to create product associations

  • The ratio between the number of transactions for a product vs. the total number of transactions for all products used to create product associations

  • How long before a recommendation expires