Maximize AI potential with the LLM Open Connector for seamless integration
Inroduction
The LLM (Large Language Model) Open Connector bridges the gap between Salesforce and cutting-edge AI models, enabling seamless integration of natural language processing capabilities into your business workflows. With this connector, organizations can harness AI to enhance customer service, automate complex processes, and provide intelligent insights—all within the Salesforce ecosystem. This article explores how the LLM Open Connector works, configuration and practical use cases to supercharge your Salesforce environment with AI-driven solutions.
Benefits of Using the Salesforce LLM Connector with a Custom Model:
- Personalized AI Solutions
Tailor the AI model to your unique business data and use cases.
- Seamless Salesforce Integration:
Quickly embed AI-driven functionalities into existing Salesforce applications without heavy coding.
- Secure & Scalable:
Utilize Salesforce’s built-in security, ensuring safe data exchanges while handling large-scale requests.
- Real-Time Insights:
Get instant AI-generated outputs, enhancing productivity and decision-making.
- Cost Efficiency:
Reduce development time and focus on business logic rather than complex integrations.
- Flexibility:
Supports both third-party models and in-house developed models, giving you full control over your AI strategy.
Hands-On Implementation: A Practical Guide to Bringing LLM Open Connector to Life
Before getting started, keep these key requirements in mind:
- You must have an LLM ready for connection, whether it's from an existing provider or a model you host yourself. It can be proprietary, open-source, or custom-built.
- The LLM Open Connector is flexible and works with any platform or model, allowing you to choose your preferred provider, inference engine, or hosting solution.
- The model must be accessible via a standard REST API endpoint and adhere to the Open Connector specification, which follows industry-standard LLM API protocols.
- The endpoint should support essential foundation model capabilities, particularly chat completions or embeddings.
Our goal was to integrate Eternity Ac's AI clones into Salesforce through the custom LLM Open Connector.
A brief look at our implementation process:
1. We start by selecting Einstein Studio Model Builder in Data Cloud, followed by clicking the Add Foundation Model button:
2. Next, we select the Connect to your LLM:
3. On the next screen, we provide key inputs, including a user-defined name, endpoint URL, private key, and model name. For this example, we’ll use the Eternity Ac model, but any LLM endpoint that follows the Open Connector API specification can be used:
4.After successfully connecting the LLM, we now have a foundation model stored in Data Cloud. Next, we click the New Configuration button to create a configured model, making it available in Prompt Builder, the Models API, and more:
5. Next, in the Model Builder Playground, we enter a simple prompt to confirm that our foundation model is functioning correctly:
After successful tests, we save our configuration with a custom name. This name will be used to identify the configured model in Prompt Builder and other areas of the user interface.
Clicking on the name of our newly configured model takes us to its details page. After reviewing the model information, we can select the Create Prompt Template button to generate a new prompt in Prompt Builder for use in Salesforce.
Creating a prompt template
- In this walkthrough, we'll take on the role of a prompt engineer and craft a basic prompt that provides answers to the questions from Case.Description and posts the results to the Case.Resolution_Summary__c field:
2. First, we switch the model type from Standard to Custom and select the Eternity Ac model we previously configured using the LLM Open Connector
In the Preview area we can test the configured prompt, by selecting a test Case and clicking the preview button we will see the result generated by our configured model. After a successful test, the prompt can be activated.
3. Here's what it looks like from the end user's point of view, the field Resolution_Summary__c has functions as an Einstein Generative AI with defined prompt template:
This feature enables users to trigger the prompt template with just one click:
Conclusion
In this post, we demonstrated how to leverage the LLM Open Connector to integrate and save a Eternity AС model endpoint in Model Builder. We then utilized it to create a custom prompt template and connected it to a GenAI text field within the core Salesforce experience. We look forward to seeing the LLMs you integrate and how you use them to enhance generative AI in Salesforce.
