> ## Documentation Index
> Fetch the complete documentation index at: https://docs.topsort.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

export const LastUpdated = ({date, lang = "en"}) => {
  const translations = {
    en: "Last updated:",
    es: "Última actualización:",
    pt: "Última atualização:",
    fr: "Dernière mise à jour:",
    de: "Zuletzt aktualisiert:"
  };
  const label = translations[lang] || translations.en;
  return <>
<style>{`
.last-updated-component {
display: inline-flex;
align-items: center;
gap: 8px;
padding: 10px 16px;
border-radius: 8px;
margin-top: 12px;
margin-bottom: 16px;
font-size: 14px;
background-color: rgba(0, 0, 0, 0.05);
border: 1px solid rgba(0, 0, 0, 0.12);
color: rgba(0, 0, 0, 0.75);
line-height: 1;
}

        .last-updated-component svg {
          flex-shrink: 0;
          vertical-align: middle;
        }

        .last-updated-component span {
          display: inline-flex !important;
          align-items: center !important;
          line-height: 1 !important;
        }

        [data-theme="dark"] .last-updated-component {
          background-color: #3a3a3a;
          border: 2px solid #888888;
          color: #ffffff;
        }

        [data-theme="dark"] .last-updated-component svg {
          stroke: #ffffff;
        }
      `}</style>
      <div className="last-updated-component">
        <svg width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">
          <circle cx="12" cy="12" r="10" />
          <polyline points="12 6 12 12 16 14" />
        </svg>
        <span>
          <strong style={{
    fontWeight: 600
  }}>{label}</strong> 
          <time dateTime={date}>{date}</time>
        </span>
      </div>
    </>;
};

T-Brain is Topsort’s commerce intelligence layer. It learns from your catalog and shopper behavior to generate predictions that help your site or app decide which products to show.

T-Brain is built on Topsort’s proprietary Large Commerce Models (LCMs), generative AI models built for commerce prediction. It is available to commerce and commerce-adjacent businesses, whether or not they use Topsort’s advertising platform.

## Commerce intelligence across your store

T-Brain can support search, homepage recommendations, notifications, custom storefronts, and ads. Retailers can use the same commerce data to train models for multiple experiences, reducing the need to build and maintain separate prediction systems for each use case.

As more models become available, retailers may be able to consolidate their technology stack around fewer vendors.

A retailer can use Quality Prediction today to help determine which products deserve a recommendation slot or which candidates should rank higher in an auction. Upcoming models will support selecting relevant products and personalizing their order across search and discovery experiences.

Existing Topsort clients can get started with the catalog and shopper behavior already synced from their marketplace.

## Models

### Quality Prediction

Quality Prediction scores how suitable a product is to show in a given context. Ranking, recommendation, and auction systems can use this score to help determine which products to display. This model is available now.

### Upcoming models

More models are coming, including **Product Retrieval**, which selects a shortlist of relevant products based on a search query, shopper, or page context, and **Product Ranking**, which reorders a list of products to show the best ones first for a particular shopper and context.

## How to use

Use the T-Brain interface to connect your data, configure training, deploy a model, and monitor its performance. Once deployed, your site or app can call the model through an API to receive predictions.

### 1. Connect your data

Open the **Data** tab to configure data syncs, including your catalog, impressions, clicks, and purchases.

If you already use Topsort, T-Brain can use the commerce data synced from your marketplace. If you do not use Topsort, configure data syncs to provide your catalog and shopper behavior.

### 2. Configure training

Open the **Tasks** tab to create a training task. Select the model you want to train, choose the dataset, and configure training.

### 3. Track training and deploy

Open the **Training** tab to view your training tasks and their status: in progress, completed, or failed.

Select a task to view its details and training timeline. You can track progress while training is running or review failure details to identify issues that need attention.

Once training is complete, open the task’s details and deploy the model.

### 4. Integrate predictions

After deployment, call the appropriate T-Brain API with an input to receive a prediction from your trained model. Your site or app can use that prediction to inform product selection and display.

### 5. Monitor performance

Open the **Monitor** tab to view the performance of deployed models. Performance data appears after a model has been successfully deployed.

See the [FAQ](/en/knowledge-base/t-brain/faq) and [data requirements](/en/knowledge-base/t-brain/data-requirements).

***

<LastUpdated date="2026-10-08" />


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.