Salesforce’s Trusted AI Layer Makes Sense After All

Salesforce CEO Marc Benioff on Tuesday held a clinic on Trusted AI and its impact on Salesforce CRM.

The idea of trust has been part and parcel of Salesforce’s approach for many years, and Benioff’s presentation added a new layer that was only sketched out previously. While I am still leery of using “trusted” as a descriptor, there is logic in the effort.

The Trusted AI Layer that Salesforce is aggressively building and deploying is significantly different from the AI and large learning model (LLM) approach that much of the industry relies on today.

According to Benioff, LLMs simply vacuum up whatever data they can capture off the internet, regardless of its provenance. But all of it is the intellectual property of someone else, which will be a problem that could bite us all in the anatomy; see below.

You might recall that one of ChatGPT’s parlor games early on was to write something in someone else’s style. Shakespearian sonnet? No problema! A CRM paper in the style of Denis Pombriant? Good luck with that.

But seriously, that’s a problem because I get paid to analyze and write, and having a generative AI product do my job could be a problem for me. Multiply that by billions of people, and what have you got?

If you work in a highly regulated industry where people’s data is sacrosanct (think finance and health care, for starters), you might care a lot, at least to the extent that you’d rather do good business than face class action suits from people whose data you used or gave away.

Precision To Enable Trust

That’s the essence of Trusted AI that Benioff and company are promoting and making happen. Salesforce’s big conference, Dreamforce, is happening in September, and it will be an important milestone for showing how Trusted AI will roll out throughout the product line.

Those products will incorporate 210 AI patents and counting, which are based on more than 220 research papers, by the way, so this is not simply a new marketing program.

So how does this square with my earlier analysis?

Yesterday I said that the word “trusted” is wide of the mark and not what people will pay for. Precision in CRM and most things that I look at is that thing. But now I believe that precision is the thing that we will next use to enable that trust. Thus, the trusted thing has become essential.

Interestingly, the AI-precision interface is the clearest demarcation of data turning into information yet. After all, it is information that makes precision possible, as in yesterday’s example of “right product, right customer, right time.”

The IT revolution of the last 50 to 60 years represents a discrete moment in the history of Western capitalism, the latest K-wave in a series that goes all the way back to Adam Smith’s “Wealth of Nations.”

For several years now, I’ve said that the era is ending. I think generative AI is putting the last nail in the IT K-wave coffin. Thanks to generative AI, IT is more critical than ever, and the end of the K-wave is not the end of the world. But AI-driven IT is shaping up to be so different that I think a few years from now, people will say this moment we’re in was a turning point.

AI Driving Economic Change, Carbon Capture

On another front, I always thought the next economic surge would come from all the effort and money we put into saving the planet.

It’s quite possible that AI will be a critical driver in that effort, especially if you consider that it consists of reducing and eventually eliminating the use of carbon fuels and reabsorbing and safely storing away some of the carbon already in the environment.

There’s a great deal of money to be made on electrifying the economy and retiring the fossil fuel paradigm. However, capturing carbon and storing it safely is costly. AI might be an ideal contributor to capturing and sequestering carbon because it can be highly autonomous and it’s virtually free.