What’s Next for Salesforce AI: A Quick Interview With Salesforce’s SVP of AI

As we inch closer to the next Salesforce release, the talk around AI remains as prevalent as ever. By now, you should be an expert on Salesforce AI in your own right – after all, it definitely had its moment in 2023!

As Salesforce work quietly in the background to create the next best tools, let’s take a look at what the next steps might be, with a bit of insight from Jayesh Govindarajan, the Senior Vice President of Salesforce AI.

What Does the SVP of Salesforce AI Do?

Having dabbled in data and engineering for a lot of his life, Jayesh actually joined Salesforce via the acquisition of MinHash in 2015 – a data science startup that Jayesh had founded. From there, he went on to become the Vice President of Data Science and Engineering at Salesforce, before becoming the Senior Vice President of Salesforce AI in 2021.

Fast forward to now, and Jayesh leads the Salesforce AI org – a pretty mean feat!

The Future of Salesforce AI

What makes a good AI strategy?

“To create good AI, you always have to have the product in mind. I like to think: how can I help?”

When it comes down to the perfect formula – that ideal balance between innovative and user-friendly – Jayesh says that it should be sophisticated, scientific, well-engineered, and familiar. “It’s all about giving [customers] something familiar, and doing something new with it,” he says. “We really try to meet customers in the middle.”

What current challenges (if any), do you face rolling out new AI products to end users and people like us?

“We don’t typically have any challenges, but like with any product, it can sometimes be difficult to foster understanding and explain value.

“But Trailhead is a wonderful resource for Salesforce AI learning – over 1 Million AI-focused badges have been achieved so far.”

This includes on Trails like Get Started with Artificial Intelligence, and with modules such as Data Fundamentals for AI and Responsible Creation of Artificial Intelligence.

How do you see the Einstein 1 Studio developing road map-wise?

When I asked Jayesh this question, his reaction was not too dissimilar to anyone’s with a good secret to hide. Nevertheless, his excitement was definitely there. “We have so many ideas.” he enthused.

He told me that they have a fairly complete step one in place (the current Einstein suite I would assume) and that everything was developing thanks to user data and movements. “As [users] build, we collect data that we use for further training and development,” he said. “This makes [the LLMs] a lot smarter.”

He also hinted that there’s a hope to branch into the frontier model space for further AI progression. This will mean using AI models that are more powerful (and cool!) than we’ve ever seen before.

“We also want to give customers more choice,” Jayesh said. This could involve expanding the existing mix-model infrastructure further, keeping relationships with old models, but providing even more options for all kinds of use cases.

Note: A frontier model refers to a class of advanced AI models that push the boundaries of what’s currently possible in artificial intelligence. These models are designed to handle complex tasks and large amounts of data, often leveraging the latest advancements in machine learning and deep learning.

Summary

Although it’s unlikely that we will receive more news on the next advancements of Salesforce’s AI tools, talking to Jayesh was a great opportunity to get a better understanding of the kind of direction that the cloud giant wants to head in.

Do you have any predictions? Let us know in the comments below!