The Truth About Salesforce's AI LLM: Flexibility is Key

Large language models (LLMs) are revolutionizing various industries, and Salesforce is at the forefront of bringing this power to the CRM landscape. But with so many options and considerations, understanding "the truth" about Salesforce's AI LLM involves recognizing its flexibility.

Demystifying the Options: Shared, Hosted, or Your Own

Salesforce doesn't force you into a single LLM solution. You can leverage:

Shared Salesforce LLMs:

  • Salesforce shared LLMs are a way to access LLMs across the internet. Salesforce pioneered this with our partner OpenAI. Salesforce uses a secure gateway to travel across the internet to ask its amazing LLMs for a response.
  • Einstein lets customers use Salesforce LLMs developed by Salesforce AI Research to power advanced capabilities, such as code generation and business process automation assistance, fundamentally transforming how businesses interact with their CRM software. Salesforce LLMs, including CodeGen, CodeT5+, and CodeTF, help companies increase productivity, bridge the talent gap, reduce the cost of implementations, and better detect incidents.

Hosted Third-Party LLMs:

  • Integrate models from industry leaders like Amazon, Anthropic, and Cohere directly within the Salesforce infrastructure. This ensures data privacy as prompts and responses remain within Salesforce.

Bring Your Own Model (BYOM):

  • Continue using your custom-trained models on external platforms like SageMaker or Vertex AI. The Einstein Trust Layer facilitates a secure connection, keeping your data within your designated trust boundaries.

Einstein 1 Platform: Simplifying the LLM Journey

The Einstein 1 Platform acts as your one-stop shop for LLM adoption. It offers:

  • Pre-built Functionality: Leverage pre-configured options to unlock LLM potential quickly.
  • Prompt Builder: Craft effective prompts tailored to your company data without model training.
  • Einstein Trust Layer: Make secure LLM API calls while safeguarding your sensitive data.

Choosing the Right Path

The optimal approach depends on your specific needs and data privacy requirements. Here are some factors to consider:

  • Use Case: Different use cases might benefit from different LLMs. Analyze your goals to identify the ideal model capabilities.
  • Data Residency: If data residency within a specific region is crucial, BYOM or hosted third-party LLMs might be preferable.
  • Existing Investments: Consider integrating your existing custom models through BYOM for a seamless workflow.

The Future of Salesforce's AI LLM

Salesforce is committed to an open ecosystem. The BYOM options are constantly evolving, so stay tuned for future updates. With Einstein 1, you can confidently explore the world of LLMs and unlock a new era of CRM innovation.