Targeted at data scientists and engineers, Einstein Studio is designed to help enterprises connect their Salesforce data to any AI or large language model, including Llama 2 and OpenAI’s GPT 4, for enhancing and accelerating the development of AI and generative AI applications, the company said.
Einstein Studio is generally available at launch and comes bundled with Data Cloud at no added cost.
Enterprises will be able to save on cost and time and be faster to market due to the in-built features of the tool, such as zero-ETL (extract, transform, load), according to analysts.
“Moving data around has been a major issue when it comes to model training as it required time-consuming integrations. Zero-ETL means keeping the data where it is,” said Andy Thurai, principal analyst at Constellation Research.
Other time-consuming issues that the tool solves, according to Thurai, include eliminating the need for data normalisation, and cleaning and providing real-time data for model training as Einstein Studio automatically “harmonises” an enterprise’s Salesforce data and draws updated data from the Data Cloud.
“Providing real-time data for model training is something that most developers are struggling with. Once you take the snapshot of the data for model development, then the latest or fresh data won’t factor in the model development,” Thurai said.
Einstein Studio also comes with other features that can help enterprises serve models and monitor them for discrepancies, the company said. The new tool can also help enterprises connect data to AI or large language models being trained on other platforms such as Amazon SageMaker and Google’s Vertex AI.
As part of the new offering, Salesforce said it will provide a control panel to help data scientists and engineers govern how their data is being exposed to chosen AI platforms for training.
The move to launch Einstein Studio, according to Thurai, helps Salesforce cater to two sets of users or customers — users who want to create and train their own custom models, and users who already have an existing custom model or want to fine-tune any open source large language model with their custom data from Salesforce.