Whether you have been advocating for or against Salesforce and Agentforce this year, you have to give the cloud giant credit where it’s due. At the beginning of this year, Salesforce announced that Agentforce had secured 8,000 (4,000 paid) deals. Cut to now, and this number has now more than doubled, with 18,500 (9,500 paid) deals.
For perhaps the first time this year, attitudes towards Agentforce are beginning to turn positively. However, nothing with Salesforce is ever clear-cut. Thoughts from inside the house have begun bleeding out, and it’s evident that the Agentforce story might have another side to it. So, what do we know?
The Good News
Last week, Salesforce announced its Q3 FY26 results. What they revealed had a very clear message: Agentforce had a purpose, and it was not just a tool that could look pretty in a demo.
In fact, Agentforce and Data 360 (formerly Data Cloud) were classed as “momentum drivers”, hitting almost $1.4B in ARR, with Salesforce boasting more than 9,500 paid Agentforce deals and 3.2 trillion tokens processed. This level of adoption signalled a 50% increase quarter-over-quarter – impressive numbers for a solution that had divided its users and its community since day one.
As my insightful colleague Henry Martin put it: “There have been warnings of a bubble in the AI sector for quite some time now, and only time will tell if artificial intelligence products like Agentforce really are going to change everything very soon – or if another, less rosy future lies in store for Salesforce, which is wagering so much on that idea.”
Agentforce’s Story is Shifting
We have covered Agentforce and its journey extensively here at Salesforce Ben, and it is safe to say that it has been far from smooth sailing. Adoption levels have been rising, but companies have consistently struggled with value realization, pricing confusion, and what they could actually do with the AI platform.
Q3, as Forbes recently put it, “marks Salesforce’s critical pivot to autonomous AI”. The initial concrete proof that SaaS can monetize AI has been shown to the world.
This also highlights the fact that Salesforce is no longer a cloud product or even a SaaS product. It was back when Sales Cloud and Service Cloud dominated their sectors; now, Salesforce is an AI company, in every sense of the term. The SaaS software is an added bonus, not the other way around.
According to Business Insider, in Salesforce’s strategic planning for this year, one action item in particular stands out: “win the enterprise agent wars”. This is no longer about just creating the next great product – it’s about solidifying Salesforce as an industry leader, much like the company has accomplished as a CRM.
Whether or not you believe CEO Marc Benioff will rename the company to Agentforce, or that these confident strides will protect Salesforce when the supposed AI bubble bursts, if you needed any more convincing that the AI train will not be slowing down any time soon, you have it.
“They Force It Down Our Throats Each and Every Day”
The prospect of winning the AI race can be enough to get the pockets of shareholders and stakeholders rustling, but for many on the ground – using and building Salesforce’s AI products – the story is very different.
Earlier this year at Salesforce’s flagship conference Dreamforce, both Salesforce and its community made themselves clear. The community was still upset over the CRM giant’s ‘Agentforce or bust’ mentality. Salesforce explained that it had to do this or risk becoming “obsolete.”
This sentiment does not appear to be unique to its users and customers either. Reports from inside the mothership have revealed issues like the inability to keep up with the changing AI vision, the products themselves being too complicated to use, and the alleged belief that some products are months or even years away from conception despite being showcased.
Kristi Valente, a Salesforce Administrator since 2009, told Business Insider that “if you are a normal business with normal admins, you do not have the expertise to set [Agentforce] up.”
“It’s very, very difficult – even for people working on the products – to know the difference between what we say in a demo, what’s on a road map, and what’s actually in production,” one senior employee at Salesforce said. “It’s a full-time job just figuring that out.”
Over on Reddit, alleged employees at Salesforce have come to say that handling such an intense push in one direction has been difficult to manage.
“Try working at the mothership where they force it down our throats each and every day,” one person wrote.
“What’s worse is the mothership pipeline has, like, an 18-24 month head start on the koolaid firehose,” another wrote.
In order to get to the top, Salesforce has to always maintain at least a modicum of ruthlessness. After all, as Co-Founder Parker Harris said during True to the Core: “AI is so complex and demands so much investment. We just have to.”
However, Salesforce’s community and Salesforce itself have come to understand that the path forward will likely not be a smooth one, with ongoing and future clashes predicted. So what can Salesforce do about it?
The Path Forward
When Salesforce caught wind that reports of their internal employees and customers struggling with Agentforce came out, they understood that the education surrounding Agentforce and the other AI products was critical.
“We have teams that are laser-focused on education, whether that’s for our Trailhead offering or internally,” Madhav Thattai, SVP and COO of Agentforce, told Salesforce Ben. “The external Trailhead training has been created from the stuff we were doing internally with our own people.”
“We do a lot of enablement at scale, bringing people on site, taking our content around the world, making sure that we have the practitioners who are really building this technology. We have our forward-deployed engineering team that has really been out there building the first of these agents with our customers directly, involved in hundreds and hundreds of implementations this year.”
For Salesforce, developing Agentforce has been broken down into two key areas: trust and design. Trust was a heavy focus last year, and this year, configuring how customers and users design with the tool has been paramount.
Through this, Salesforce has been able to identify where users are struggling most: data context, determinism, and sculpting the perfect UX layer.
For Madhav, this understanding has been crucial, as it has allowed his team to help in the areas where help is needed most, rather than just send people to Trailhead or offer general advice.
“We’re in there with them helping them be successful, and we’re learning a lot about the product along the way,” he explained. “That’s also an opportunity to continuously train our people and our customers’ people on how to build these agents.”
Final Thoughts
Even before any of us knew whether Salesforce’s Q3 earnings would be positive or negative, the belief that the division between the company and its people would still persevere was one that did not seem too far-fetched. A week later, and this belief already seems to be materializing.
Salesforce’s Agentforce situation can feel a bit one step forward, two steps back at points, but what is crucial now is letting the data speak for itself. Agentforce is succeeding and growing, and it is here to stay. An emphasis should always be placed on core functionality and getting the basics of a platform right – both aspects that Salesforce has appeared receptive to assisting with – but crunch time is here.
Salesforce is no longer simply a cloud company or a SaaS company, and it likely never will be again.