This is the moment for truth for Salesforce. After nearly a year and a half in the world, Agentforce – the apple of Salesforce’s eye – has progressed in leaps and bounds, but how many people are actually using it?
For much of its short lifetime, Agentforce has been regarded as a flashy product that companies keep their eyes on until the right use case comes up or the investment makes sense. However, Salesforce’s end-of-fiscal-year reporting is doing a lot of heavy lifting to convince investors that Agentforce is the hottest tool in the tech stack.
The Exponential Growth of Agentforce
Before this week, the latest Agentforce numbers publicly disclosed came in January, with Salesforce CEO Marc Benioff taking to X to reveal that Agentforce had hit 11.14 trillion tokens served.
Salesforce’s Q3 financial results, which came out in December 2025, revealed that Salesforce had closed 18,500 Agentforce deals, 9,500 of which were paid, with 3.2 trillion tokens processed. This signifies that in just one month, token usage had increased by a staggering 285% according to Benioff.
As my colleague Henry Martin covered recently, this number of tokens processed strongly indicates that a notable number of customers are actively using Agentforce, but this only paints the picture of customers who have Agentforce in their tech stack already. The sentiment has continuously been that most of Salesforce’s customers are not using Agentforce, but those who are using it are dedicated to getting the most out of it.
With this in mind, Salesforce has stressed that Agentforce remains the CRM company’s “fastest-growing organic product in Salesforce history” with a growth rate of 50% QoQ.
“These aren’t vanity metrics,” a Salesforce spokesperson told SF Ben. “We’re seeing record growth, thousands of new paid global customers, and a surge in production deployments.”
Lessons Learned
Although his positive outlook on Agentforce illustrates Salesforce’s confidence in the technology, it has not come without trials and tribulations.
At the beginning of last year, community opinions were mixed as Salesforce navigated the product’s infancy. Some professionals saw the value in what it could do right away, while others criticized the confusing pricing, setup, or lack of tangible use cases.
Salesforce also recently went back to the drawing board as it assessed the role LLMs played in Agentforce’s development, admitting that it had “more trust in the LLM a year ago” and was now taking on a more deterministic approach to Agentforce automation.
Agentforce’s product team has also had a shakeup, with former EVP and GM of AI, Adam Evans, stepping down and Madhav Thattai, the former COO of Agentforce, stepping in to replace him.
We now know that Salesforce has big plans to bring Agentforce and Slack closer together, providing insight into the direction Salesforce will be heading in the coming months.
Nearly 30,000 Agentforce Deals
On Wednesday, Salesforce announced the latest Agentforce adoption numbers, with a total of 29,000 closed deals since the product’s launch. This is up 50% QoQ, Agentforce ARR reaching $800M.
Henry highlighted that the growth rate of 50% is particularly significant, as it could mean that if this rate continues, we could see Agentforce cross the $1B ARR threshold this coming quarter, standing at $1.2B in Q1.
Salesforce has also introduced Agentic Work Units (AWUs) – a measure of discrete tasks executed by AI agents in production. So far, the SaaS leader has delivered 2.4B AWUs, growing 57% QoQ.
This finally provides clarity on what many current and prospective Agentforce customers have been looking for: tangible value realisation. CIOs can now compare AWUs-per-dollar against fully-loaded employee cost, effectively helping them understand how valuable an employee is versus an AI agent.
“AI Is Now Causing the Software Sector to Be Revaluated”
Vernon Keenan, Senior Industry Analyst and founder of Keenan Vision, said that he believes that AI has now triggered a revaluation effort within the software sector, as part of a wider “secular transition”.
“It seems like there’s a ‘baby in the bath water’ phenomenon going on here,” he told SF Ben.“I don’t see it as a cycle necessarily. I think that AI is now causing the software sector to be reevaluated, and since it’s so heavily dependent on SaaS and people haven’t figured out the business models yet, that’s why all the [stock] prices are going down.”
This adds up – despite what Salesforce called a “record fourth quarter”, the CRM stock still dipped in extended trading as it forecasted revenue below Wall Street estimates.
Enterprise-Wide Adoption Will be Led by Employee Agents
Mehmet Orun, Salesforce MVP and Data 360 expert, says that he predicts this will be a bigger year for Salesforce and AI as methodologies mature.
“I expect Salesforce to emphasize customer-facing agents,” he told SF Ben. With consumer agents, the ROI is very tangible – they drive token consumption, and you can easily measure business value. However, this also means organizations of all sizes are expected to trust AI with their mountains of data, which holds consumer agents back. This is where employee agents come in.
He explained that many organizations are following MVP best practices and starting with small use cases, often leveraging only the data in their CRM org. “However”, Mehmet highlighted, “the greater value potential, powered by Data 360, lies in tapping into relevant data assets across the organization to create contextual intelligence – where humans and agents alike have access to information that is relevant, complete, current, consistent, correct, and compliant.
“I expect this year to bring more AI and Data conversations – not just feature discussions.”
Mehmet also highlighted that if Salesforce wants Agentforce to accelerate, greater emphasis must be placed on data reliability.
Market interest in agents appears to be growing, and at a quite considerable rate. However, most companies are in a different place in reality compared to where they want to be, and Mehmet stressed that before any company goes all-in on AI, “data quality cannot be an afterthought.”
He pointed to a recent Salesforce-led webinar where data reliability was not listed among the core AI adoption success factors – highlighting an opportunity to elevate the conversation.
He added that he expects the Salesforce community – sharing best practices and anti-patterns, and the broader AppExchange ecosystem – continuing to extend features, assess and address data quality issues, to play an increasingly important role in bridging these gaps and helping organizations build more resilient data foundations for AI.
Is Agentforce Confidence Slowly on the Rise?
Although it is clear that a large proportion of the Salesforce ecosystem believes that there is still more work to be done with Agentforce, are the tides finally shifting in the right direction?
“Having seen every hype cycle – both as a customer for 12 years and an employee for 10 – and having interviewed over 1,650 senior business leaders on my podcast, I can confidently say: this moment is different,” wrote Salesforce’s Chief Digital Evangelist Vala Afshar on LinkedIn. “Ultimately, the fittest businesses are not defined by size or speed, but by the ability to adapt. This speed of deployment is nearly unprecedented.”
Ian Gotts, the Founder of Elements.cloud, said that Salesforce’s latest results should quell some hesitation around Agentforce, but that we are still in the early stages of agentic development.
“These results should steady some nerves about Salesforce, but we are in the very early days of agentification,” he wrote on LinkedIn. “Organizations are facing up to the dramatic transformations they are going to need to make. Salesforce needs to ensure that it is in its strategic plans both as the core data, but also the agent platform.”
Community outlook tells one side of this complex story. Many professionals in the ecosystem still haven’t got round to using Salesforce, and those who have certainly have mixed opinions. However, Agentforce’s latest expansion numbers have shown that over 60% of Agentforce and Data 360 bookings in Q4 came from existing customer expansion – likely indicating that those who can get stuck in with the tool want to get as much from it as they can.
Final Thoughts
With the financial forecasting and Salesforce’s Agentforce numbers telling different sides of the same story, the CRM giant’s proprietary AI tool does not seem to be in the clear just yet, but perhaps this is to be expected.
Salesforce is evidently making moves with Agentforce, and this can be seen in the adoption figures and closed deals. The community is paying attention, but more time is needed to see if it’s actually being utilized in the way that Salesforce needs it to be.
