This YouTube panel (Salesforce Research leaders Silvia Savarese, Nikhil Naik, and Shelby) describes Salesforce’s AI research and how it drives product innovation. They emphasize building and deploying large language models and recommendation systems to improve sales, service, and developer productivity, while stressing trust, brand-aware generation, and efficiency (carbon/power reduction). They also discuss societally impactful projects (protein generation for novel molecules, computer vision for breast cancer receptor status) and future directions including autonomous agents and personal AI assistants.
Chapters
The future AI is about making users and customers more creative to provide much better experience.
How do we push this area of AI forward? But always in the back of our head we're thinking, how would this fit in the product?
We've been developing groundbreaking applications for sales, service, marketing, CRM.
Not about replacing humans, but really about empower humans with new capabilities.
What we also worked on is how we can take the same AI and apply it to impactful applications for society.
We've been defining the frontier of AI for years. But to stay ahead, you have to be looking ahead.
I'm Silvia Savarese, I'm a chief scientist at Salesforce.
My name is Nikhil Naik. I'm a director of AI at Salesforce Research.
I'm Shelby and I lead a team of AI researchers.
Salesforce has been performing groundbreaking research in natural language processing, computer vision, reinforcement learning applications for many years now. We have built more than half a dozen LLMs. We have been working on large language models since about 2018 and we have worked on every iteration of large language modeling technology since then.
AI and trust go hand to hand. So we are committed to create an AI can be trusted by our customers. So one of the key priorities we have in our organization.
Salesforce as a company has a wide variety of applications for AI and enterprise and is one of the best places where you can take AI technology and empower thousands of customers. At Salesforce, we have invested a lot in developing large language models for developer experience. So these are models that can generate and correct and edit code and that can help developers write high quality code.
And this type of technology is currently in the process of being deployed to customers for general usage.
This can help them become more productive while using Salesforce applications. We worked with a customer on applying AI for service application to tell a better story about their product to their customer. We were able to build a large language model that that helped customer service agents tell better, more brand aligned and more interesting stories about their products to their customer in real time.
We've trained the model to be able to generate replies that are in the voice of the brand that we're working with. So it does give the product information to the service agent so they're able to deliver the best customer service, but it also delivers that brand, that essence of the company.
And this was an exciting application of using generative AI and also an example for us where working collaboratively with the customer leads to a novel application that we may not have thought about before.
Recommendation Systems is a type of AI model that given data can make a recommendation for products. What we've done is built recommendation systems to help our account executives. For example, we've built a recommendation system to recommend apps for them to sell. So this really helps with customer engagement and this saves the AE time.
We are committed to use AI technology to also have an impact on sustainability and environment.
AI Salesforce is investing in in house AI models that are very efficient to run and that can reduce carbon emissions and reduce power usage. When our customers use our generative AI applications.
By making operations more efficient, more effective, by reducing cost to serve, and by reducing compute, we can have a direct impact to the way this technology can be deployed. Exciting element of our work is that not only we are impacting the business of the company, but also having a larger societal impact.
What we have done is we have taken large language models like ChatGPT and applied them to the problem of generating novel protein molecules. And this technology can be used for developing new medicines, for developing new sustainable chemicals, and for developing other groundbreaking biomolecules which can have lots of applications. Salesforce Research has been working on computer vision AI for a long time, which has applications in things like E commerce and marketing for understanding and predicting attributes of images. Salesforce Research took the same technology for predicting hormone receptor status of breast cancer images which can help improve therapy decisions for breast cancer patients.
To me, the future of AI is autonomous agents. An autonomous agent can make plans and even execute plans. And we're working so hard to get AI agents to our customers. Let me break down what an AI agent really is. We've all worked with chatbots. Before. You go to any website, there's a chatbot there to help you. I need help with so and so the chatbot gives you a response. So that is usually backed by a large language model. But we're thinking about the next generation, what comes after that. And that is what an AI agent is. It can plan the actions to take and it can actually take those actions. So this is extremely powerful. So imagine you're using a chatbot. You want to plan a trip to Mexico. So not only will the chatbot tell you in text what to do, it will actually take the actions for you to. So it could actually go to the websites and book the tickets for you.
I like to see a scenario where each user, each customer will have their own personal assistant.
A lot of manual clicking or repetitive actions. I think that's all going to be gone. A lot of that's going to be automated.
To me, the future of AI is all about enabling people to be more productive, healthier, and happier.
Customers are going to be able to achieve so much more in very little time on our Salesforce platform. It's exciting. It's like every day there's a new way we can apply AI AI is.
Ushering in a new era of productivity. With decades of history in an enterprise world, there is no better place to pursue this vision. At Salesforce.
