Salesforce Hints at These Major AI Updates in 2026

The pressure on the artificial intelligence market has reached new heights, and we’ve been hearing something distinct between customers, investors, and analysts alike: “We want more.” More innovation, more capabilities, and more access. 

For Salesforce, this year is a chance to meet their customers and stakeholders where they are: at the forefront of groundbreaking technology that they might not be fully convinced by. So how does the CRM giant do this, and what can we expect from it this year?

What Are the Top AI Predictions for 2026?

It perhaps goes without saying, but 2026 is expected to be a colossal year for artificial intelligence.

The AI 2027 report – an encyclopedic report written by a handful of the most prolific names in AI – details that both AI development and scale will grow exponentially this year. One main prediction is that using AI to train AI will pay off, which is something that Salesforce has been doing since the end of last year, and possibly before that, too. 

Although not a direct link, Salesforce has also experimented with this methodology through eVerse, an agent simulation environment, where AI agents can learn from other AI and improve over time.

Come mid-2026, and predictions are centered around the growing AGI race. Unsurprisingly, the battle is likely to be bigger than just the company versus company dynamic, with individual countries – predominantly the US and China – battling it out. 

Last year, Salesforce CEO Marc Benioff insisted that AGI (artificial general intelligence) was very likely to appear in the future, but that they were not part of the “state of technology today.” Could AGI be more than a vision in 2026? The odds are low, but the chances of work being done on AGI in the background this year are certainly much higher.

As we near the end of the year, AI 2027 predicts that AI will start taking jobs. This is interesting to hear, especially if you kept up with layoffs last year and just how many were already attributed to artificial intelligence in some way.

Last year, I discussed how there was a strong likelihood that AI was being used as a scapegoat or excuse for a notable number of layoffs. This remains a steadfast belief of mine, especially as new data from Oxford Economics has revealed that AI-related layoffs made up just 4.5% of total reported layoffs last year.  

However, this does not denote the possibility of AI-driven layoffs this year. In fact, Goldman Sachs has also warned that 2026 could be quite the year for it, especially as businesses experience a heightened desire to permanently lower employee-related costs.

Our 2026 AI Predictions

Here at SF Ben, our bets are placed firmly on Agentforce’s success, but if Salesforce doesn’t deliver, the outcome could be detrimental. 

My colleague Henry Martin wrote that even if Agentforce were to experience “mediocre” adoption this year, it would do wonders for investors, putting faith back in the stock, the technology, and the direction of Salesforce as a whole. 

If the scenario doesn’t play out this way, however, we could see the bursting (or at the very least the deflating) of the supposed AI bubble and a catastrophic decline of trust and perception. Crunch time is now.

Salesforce’s 2026 AI Goals

Although Salesforce is notoriously quiet about its upcoming releases, products, and enhancements, it was able to provide more information on the predominant focus of the company in the near future. 

As agent adoption continues to increase, we will find more agents across our tech stacks, in our workflows, and in our teams. The IDC predicts that total active agents will surpass 1 billion by 2029 – which is a monumental milestone – but it poses the question: what use will all those agents be if they’re just… out there? 

Agent-to-agent connectivity and cohesion are becoming increasingly important topics, which is why a current focus for Salesforce is multi-agent frameworks.

“How Do You Manage a Multi-Agent Workforce?”

Salesforce recently told SF Ben that multi-agent capabilities were on the way, allowing users to manage a multi-vendor, multi-agent workforce. The updates will focus on ensuring a unified way to govern, orchestrate, and manage agents from different companies like Workday, Salesforce, and ServiceNow.

Eddie McGraw, the VP of Corporate Communications at Salesforce, claimed that the most popular question Salesforce receives at its CIO customer advisory board meetings is how to set up and manage a multi-agent workforce.

“Nobody is doing robust agent-to-agent activities yet,” he said. “There isn’t really one unified way to manage, govern, and orchestrate them, and we’ll probably be talking about that more later.”

He highlighted that this is something that MuleSoft was already beginning to tap into with its agent fabric, which was recently expanded. 

“MuleSoft for so long maybe wasn’t a headline grabber,” he said. “Now you’re kind of realizing that Mulesoft’s had this value for years, and it’s not like it’s a net new capability for MuleSoft because they’ve always been about integration and APIs and connecting different systems.

“It’s basically just like now, let’s add this thing in and bring agents into the mix, and make sure that they’re governed appropriately.”

Final Thoughts

This is definitely not the last you will hear about multi-agent capabilities in Salesforce – so watch this space – and we cannot wait to hear about what other updates Salesforce has up its sleeve. 

In terms of predictions, further news on Agentforce adoption, AGI, and further AI-attributed layoffs have yet to come out, meaning that we need to wait a while longer to make those comparisons. Which do you think will be addressed first?