Salesforce Inc. used its messaging over the past week to elaborate on plans for delivery of its artificial intelligence agents. Yet beyond the hoopla surrounding the rollout of the company’s Agentforce platform can be found an ongoing conversation within the tech industry around exactly where the AI train is taking us.
In presentations from leading tech luminaries and private interviews with Salesforce executives during the company’s annual Dreamforce gathering in San Francisco this week, a picture emerged of fundamental changes underway for society, work and the technology industry itself. The growing use of agents, intelligent pieces of software designed to perform specific tasks, could propel the global use of AI to a new level.
“Our industry has been an industry of tools, but for the very first time, this is going to be an industry of skills,” Jensen Huang (pictured), chief executive of Nvidia Corp., said during an appearance at the conference on Tuesday afternoon. “We’re going to have agents that are excellent at particular skills. The progress you are going to have with agents over the next year or two is going to be phenomenal and surprising.”
AMD’s Lisa Su spoke with Salesforce’s Marc Benioff on Wednesday.
Huang’s belief in the future of AI agents echoed a theme expressed by many during the Salesforce event. Since OpenAI’s ChatGPT was released for general use near the end of 2022, businesses large and small have been pursuing a path toward finding a way to realize a return on their AI investments. Not all have been successful, but this is an age of experimentation according to one prominent tech executive, and it will not be the “bubble” that some have predicted.
“This is not a fad,” Lisa Su, chair and chief executive of Advanced Micro Devices Inc., said in conversation with Salesforce CEO Marc Benioff during the conference on Wednesday. “This is actually the next wave of what you can use computing for.”
Building agents on the Salesforce platform
As demonstrated by the lines of Dreamforce attendees this week in areas of the venue set aside for building AI agents, customer relationship management practitioners have no shortage of ideas around how to employ AI. Based on conversations with attendees during the week, Salesforce users were interested in leveraging out-of-the-box solutions in low-code or no-code environments to build agents that could analyze data, comprehend customer needs and take appropriate action.
Dreamforce 2024 attendees lined up to build prototype AI agents.
After entering a small amount of information, and then defining the task using a PC-connected microphone that immediately transcribed language into text, many users needed only 30 minutes to generate their very own AI agent in prototype to take back and show colleagues at home.
For Salesforce, the response from its users and comments from top semiconductor executives Huang and Su represented a validation of its strategy to overlay Agentforce as a overarching rainbow across CRM solutions such as Customer 360 and Data Cloud. Rather than offer a standalone AI product, Salesforce believes that it can differentiate itself in the marketplace by leveraging the core elements within its existing platform.
“We’re at this inflection point where all the technologies we’ve been building have prepared us for this,” said Sanjna Parulekar, vice president of product marketing at Salesforce, told SiliconANGLE. “It showcases that we are a platform company, we are at our heart a platform.”
Tracking startups and open source
As it deploys Agentforce across its platform, Salesforce is also paying close attention to the startup community that is forming around AI. The company’s investment arm, Salesforce Ventures, has demonstrated a series of savvy bets over the years with investments in established companies such as Snowflake Inc., Databricks Inc. and Zoom Video Communications Inc.
In a briefing for the media at Dreamforce on Wednesday, Claudine Emeott, a partner in Salesforce Ventures, spoke about the firm’s recent investment in Uplimit Inc., an AI enterprise learning company, after their colleagues at Databricks saw potential in the platform to educate its own customers.
“We see AI as an amazing way to catalyze change at a faster pace,” Emeott said. “This shows the pull that is happening.”
Pull is also happening in the AI open-source world, as described by Vipul Prakash, co-founder and CEO of Together AI in the media briefing with Emeott. Salesforce Ventures led the recent financing round for Together AI, which started as an open-source platform for AI workloads and now offers its own development tools for embedding AI into business applications.
Prakash told the assembled media that interest in open-source AI models will shape the market for enterprise use and adoption. “We think there is a big branch of AI that will be driven by open-source models,” Prakash said. “You are seeing ecosystems growing around these models. It lets you control the cost for companies that are productizing them.”
The agent-based focus of Salesforce this week highlighted how the flywheel of AI is currently reshaping the computer industry narrative. Software remains a key part of the conversation. Yet, as Nvidia’s Huang noted, it will be worth watching how the adoption of AI agents changes the systems and platforms that will power this next chapter of enterprise architecture.
“There’s this belief that AI is this giant wad of software that’s pulsing and growing,” Huang said. “New AIs are helping us create new computer systems… it’s a system of AIs. Nobody should miss the next decade of technology advancement. You’re not going to want to miss this moment.”