Forget the copilot: Salesforce is embracing the AI ​​agent. This iteration of AI is set to be a new wave of innovation and will be ubiquitous. However, the idea of ​​AI agents predates Dreamforce and is certainly not unique to Salesforce. How important are the agents, really?

The era of the AI ​​agent has dawned. This is not only said by Salesforce CEO Marc Benioff, but also by Sam Altman of OpenAI. Altman recently described “level 3 agents” that can reason and that can be trusted with a certain degree of autonomy. The AI ​​guru’s description fits perfectly with the agents Benioff has in mind.

Previously, Google launched the Vertex AI Agent Builder, which allows companies to create AI agents, and Microsoft is also a strong believer in AI agents. Anyone listening to Benioff and Clara Shih, CEO of Salesforce AI, during Dreamforce would almost forget that the idea of ​​the AI ​​agent is not new. Agentforce is presented with almost aggressive enthusiasm and seems to be the only solution.

Trumps and confusion

Salesforce has its trump cards up its sleeve. Agentforce integrates with the existing Salesforce environment, can access company data via Data Lake, and is compatible with existing flows and APIs. In other words, with Agentforce, Salesforce is lowering the barrier to introducing an AI agent.

The world has been confused for the past eighteen months

Clara Shih, CEO Salesforce AI

Shih likes to go a little further. In her opinion, Salesforce is the only company that has all the building blocks to introduce a full-fledged agent. The rest of the world needs to be educated. “The world has been confused for the last 18 months,” she says. “People thought that they could easily integrate such a model into an enterprise context because they could just use ChatGPT. But that’s not how it works.”

Feeling of exaggeration

But nobody really thought that. Microsoft, OpenAI and Google have launched copilots that can support employees based on company data. It proved difficult to get an AI copilot to work within defined limits. For example, the Microsoft copilot contained a bug that made sensitive data accessible.

It’s painful, but also somewhat unavoidable with new technology. Who says Salesforce agents will never have a vulnerability? Disney, who Marc Benioff says he’s a huge fan of, recently announced it’s abandoning Slack due to security concerns. Salesforce is in trouble, too.

This other data cloud…

Snowflake also covered AI in depth during its conference. There we again saw systems for business use where models based on business data provided correct answers in a reliable and secure way. Snowflake even helped develop its own efficient model Retrieval extended generation (RAG), which at first glance has some similarities with the Atlas Reasoning Engine which powers Agentforce.

Snowflake also talks about cornerstones for enterprise AI. According to founder Benoit Dageville, you shouldn’t go to Salesforce for that, of course.

Cards on the table

Many cards were already on the table for Dreamforce 2024:

• It Idea of ​​autonomous agents has been around for some time. Microsoft, Google, and OpenAI were already working on it, as was Salesforce. We’ve even seen early implementations like the Vertex AI Agent Builder.
• It Importance of security was never up for discussion. Employees may have thought they could feed corporate data into ChatGPT, but companies large and small are working on ways to securely integrate AI.
• The correct data are essential, Snowflake told us a few months before Salesforce at the same Moscone Center in San Francisco. Like Salesforce, Snowflake said the solution was to build AI on top of the Data Cloud. But Snowflake had a data cloud for Salesforce.

So Agentforce is not as revolutionary as Benioff-Ohana would have everyone believe. On the other hand, Agentforce is certainly more than just a marketing term. Salesforce basically offers something that no one else can currently offer.

iPhone moment?

Customers who have fully adopted the Salesforce platform have the foundation for agents. No other party can currently offer a comparable out-of-the-box experience that is even remotely SMB-friendly. Salesforce owns the data, the automations, the internal and external channels, and the API integrations, and can now click agents between them. That’s powerful.

Because inventions that change the world are rarely real inventions. Steve Jobs did not invent the smartphone: there were many smartphones before the iPhone. In fact, the PDA dates back to the 1990s. However, Apple has brought together all the relevant components into a user-friendly whole.

In fact, Salesforce is now doing the same thing with agents. Are we seeing Benioff’s iPhone moment? In any case, Shih believes that agents will be as ubiquitous as apps and websites in the coming years.

Not for everyone

However, Salesforce lacks one thing: accessibility. In 2007, anyone could buy an iPhone. Owning a Mac computer was not a requirement. That’s not the case with Agentforce. If you want to use Salesforce’s convenient, quickly deployable and secure agents, you need Salesforce Foundations. You must be a customer and have your data in the Salesforce Platform and Salesforce Data Cloud, otherwise you can’t get started.

Salesforce may be able to accelerate agent adoption. The Agent Force announcement is particularly relevant for those already familiar with the Salesforce story, but the competition will still force you to think. How can companies outside of Salesforce get started with agents quickly and securely? And how do you democratize AI agents for companies large and small? Can you even integrate AI agents securely without committing to a single ecosystem?

Benioff and Shih posed this challenge to Microsoft, OpenAI and the others. Talking about and experimenting with agents is one thing, deploying them at scale is another. Even if that scale is limited to their own customers, the competition still needs to find a good answer.