With AI upgrade, Salesforce's Einstein Copilot will handle unstructured data
With the term "copilot" gaining popularity in the generative AI space, it may soon achieve the same ubiquity as terms like ChatGPT and AI chatbot. And now Salesforce is giving its own copilot a leg up in this competitive arena.
At the Salesforce World Tour New York 2023 event last week, Salesforce unveiled an upgrade for its Einstein Copilot, a conversational AI assistant to be integrated into all Salesforce applications. Einstein Copilot was first announced in September and is launching in February 2024.
Also: How to improve your privacy on Google Bard with this one simple setting
With this upgrade, Einstein Copilot will be able to retrieve information from unstructured data, which refers to data not formatted as an organized data entry, including materials such as PDFs and emails. This feature should prove popular with Salesforce customers -- including sales, customer service, marketing, commerce, and IT professionals -- who can benefit from optimizing everyday business operations -- email, for example -- that are often not neatly organized into datasets.
Salesforce also unveiled Einstein Copilot Search, which will be found in Einstein Copilot and have "enhanced AI search capabilities" to answer complex prompts and provide smart suggestions by tapping into real-time unstructured and structured business data.
Einstein Copilot and Copilot Search will be capable of accessing unstructured data by leveraging Salesforce's Data Cloud Vector Database that unifies all business data, including unstructured data, such as transcripts and documents, and structured data, such as product inventory or purchase history.
Also: Is prompt engineer displacing data scientist as the 'sexiest job of the 21st century'?
Another benefit of the Data Cloud Vector is that it will circumvent the need to fine-tune large language models (LLMs), thereby saving businesses time and money and giving LLMs access to information that used to be unattainable due to training data limitations, according to Salesforce.
"The Data Cloud Vector Database relieves the challenge of costly and complex processes to harness the value of unstructured data," said Rahul Auradkar, Salesforce EVP and GM of Data Cloud and Einstein. "Now, our customers can reason over the full spectrum of their enterprise data to power their business applications more effectively. "
Salesforce's Data Cloud Vector Database and Einstein Copilot Search will be in pilot in February 2024, while Einstein Copilot will be generally available at that date.
Also: Generative AI filled us with wonder in 2023 - but all magic comes with a price
Although Salesforce's Copilot has not been released yet, Microsoft has many different Copilots for different enterprise needs that give users a good idea of how Salesforce's will function.