Salesforce Einstein Studio’s Model Builder is a robust tool that empowers users to create, train, and deploy custom AI models using their own datasets. Designed to streamline predictive and prescriptive analytics, it seamlessly integrates with Salesforce applications and other data sources. Model Builder enables businesses to bring existing predictive models to the Einstein 1 Platform, connect their large language models (LLMs) to the Einstein Trust Layer, and manage all models efficiently from a unified control panel.
Key Features of Model Builder
- No-Code/Low-Code Interface
Model Builder provides an intuitive interface for users without extensive data science or machine learning experience. It allows users to define objectives, configure model settings, and deploy solutions with minimal coding.
- Prebuilt Connectors
The tool integrates effortlessly with Salesforce CRM data, Snowflake, AWS, and other external data sources.
- Automated Model Training
Einstein Studio leverages automated machine learning (AutoML) to streamline the training process. Users can select the type of model they need—classification, regression, or recommendation—and the system optimizes the process by testing multiple algorithms and hyperparameters.
- Customizable Models
Advanced users can fine-tune models and adjust parameters, ensuring the model aligns with specific business goals.
- Performance Monitoring and Feedback Loops
After deployment, Einstein Studio offers monitoring tools to track model performance and improve accuracy over time. Users can retrain models with updated data, ensuring that predictions remain relevant as business conditions evolve.
- Built-In Security and Compliance
Data privacy and compliance are prioritized in Model Builder. It adheres to Salesforce’s robust security framework, ensuring customer data remains secure and complies with regulations.
How Model builder differs from Prompt builder
Prompt builder is grounded in Salesforce data whereas Model builder is not.
How to create a model using Model builder
- Navigate to the App Launcher in Salesforce (the grid icon in the top-left corner).
- Search for and open Einstein Studio.
- Click ‘New’ to start creating a model
- There are two main options for building your model:
- From Scratch: Build and train a new predictive model directly in Salesforce.
- Bring an Existing Model: Import a model created outside Salesforce via API integration.
To learn the steps involved, click Create a Model From Scratch.
- The ‘Foundation Models’ tab in Einstein Studio offers a list of pre-existing models that you can use or customize.
- You can integrate your own large language model (LLM) by selecting ‘Add Foundation Model.’ To connect your custom AI model, simply provide the endpoint details and the access key.
- If you want to modify an available Foundation Model, you can do so by creating a variation. Select the model you want to customize, then use the Advanced Settings to configure details like the number of unique keywords the model should generate in its responses.
- Each Foundation Model displays the number of instances or variations currently available for use.
- Once a model is enabled, it becomes accessible across all Salesforce channels, ready for seamless integration into workflows and applications.
- You can design custom prompt templates using these AI models to tailor the responses and outputs to your specific needs.
Take Five Consulting is a technology company, based in Virginia U.S., that specializes in the Mortgage Banking vertical especially LOS implementation and application development. Take Five Consulting creates and implement mortgage technology and software specifically for Mortgage Industry.
