The Workforce That Never Clocks Out
Agentforce is Salesforce's bet that the next hire at your company won't fill out a W-4. Autonomous AI agents, priced like labor and measured in work units, are now a $1.2 billion business - the fastest-growing product Salesforce has ever shipped.
Two dollars a conversation. When Salesforce put that price tag on Agentforce in the fall of 2024, the number raised eyebrows across the ecosystem - software is priced by the seat, not by the chat. But the tag did something clarifying: it told every buyer exactly what they were purchasing. Not a license. Not a feature. Labor. Fifteen months later, that framing had found its way into roughly 29,000 deals.
Today, Agentforce is the fastest-growing product in Salesforce's history. It closed fiscal 2026 at an $800 million annual revenue run rate - up 169% year over year - and by April 2026 the figure had climbed to $1.2 billion. In a single quarter, the platform processed 771 million of what Salesforce calls Agentic Work Units. The vocabulary is the tell. Nobody counts work units for a chatbot.
From suggestion to execution
The pitch, delivered by Marc Benioff to 45,000 attendees at Dreamforce on September 17, 2024, was that enterprise AI had been stuck in the copilot era: systems that suggest, summarize, and then wait politely for a human to click approve. Agentforce agents skip the waiting. They resolve support cases, qualify leads, create service tickets, and recommend products - autonomously, grounded in a company's own Salesforce data. Salesforce calls this the Third Wave of AI. General availability came just 38 days after the keynote, on October 25, 2024.
The brain of the operation is the Atlas Reasoning Engine, a proprietary system built to simulate how humans plan: break the task down, gather the relevant data, decide, act, check the result. It is the difference between a parrot with good manners and a colleague with a to-do list. Atlas is what lets an agent handle a refund request end to end rather than drafting a sympathetic email about it.
The proof is in the queue
Keynotes are cheap. Ticket queues are honest. At Wiley, the 219-year-old publisher, Agentforce lifted self-service resolution and efficiency by over 40%, helped onboard seasonal support agents 50% faster, and delivered a 213% return on investment - about $230,000 in savings. There is a short film about it on YouTube, which is a sentence rarely written about customer support tooling.
At OpenTable, an Agentforce agent took over the restaurant support queue and, within three weeks, was handling 73% of all restaurant web queries - a 50% improvement over the previous tool - while autonomously creating service tickets for the cases it could not close. At Saks, the agents work the other side of the counter, surfacing personalized product recommendations so human reps can spend their time as stylists rather than search engines.
What the agents actually did
Eating the dog food, saving the hours
Salesforce ran the experiment on itself first. Internally, Agentforce has handled more than 2.8 million interactions, and the company estimates its employees have saved over 500,000 hours through Agentforce in Slack alone. The cobbler's children, for once, got the shoes first - and then the cobbler put the receipts in the earnings call.
The expansion pattern suggests the pilots stuck. More than 60% of Agentforce and Data 360 bookings in Q4 of fiscal 2026 came from existing customers scaling up rather than new logos kicking tires. In enterprise software, the renewal with a bigger scope is the only product review that matters.
From 1.0 to 360
Agentforce 360, generally available since February 2026, extends the platform from answering questions to running workflows - and folds in voice, so the agent can pick up the phone. Combined with Data 360 (which now incorporates Informatica), the annual recurring revenue of Salesforce's agent-and-data stack passed $2.9 billion at fiscal year end. Enterprises in 124 countries are deploying agents, with escalation rates to humans as low as 5%.
The controversial $2-per-conversation price has evolved, as prices do. But the idea it introduced has hardened into strategy: stop selling seats, start selling finished work. Whether the phrase "digital labor" ages as a prophecy or a slogan, the queue-level numbers - 73% at OpenTable, 213% at Wiley, half a million hours inside Salesforce itself - are already in the ledger.
Vital statistics
- Launched at Dreamforce on September 17, 2024; generally available October 25, 2024
- Powered by the Atlas Reasoning Engine, built to plan tasks the way humans do
- Initial pricing: $2 per conversation, with standard volume discounts
- Part of roughly 29,000 deals in its first fifteen months on the market
- $800M ARR at FY26 close; $1.2B by April 2026 - fastest-growing product in company history
- 2.8M+ internal interactions; 500,000+ employee hours saved via Agentforce in Slack
- Agentforce 360 reached general availability on February 23, 2026
Frequently asked
What is Salesforce Agentforce?
Agentforce is Salesforce's platform for building and deploying autonomous AI agents that handle tasks across service, sales, marketing, and commerce - grounded in a company's own Salesforce data and powered by the Atlas Reasoning Engine.
When did Agentforce launch?
It was unveiled at Dreamforce on September 17, 2024, and became generally available on October 25, 2024. Agentforce 360 reached general availability on February 23, 2026.
How much does Agentforce cost?
Initial pricing started at $2 per conversation with standard volume discounts. The pricing model has evolved along with the product since launch.
How is it different from a chatbot or copilot?
Copilots suggest actions for humans to approve; Agentforce agents autonomously plan and complete tasks - resolving cases, creating tickets, qualifying leads - using the Atlas Reasoning Engine to reason over live business data.
Which companies use it, and what results have they seen?
Wiley boosted self-service and efficiency by over 40% with a 213% ROI; OpenTable's agent handled 73% of restaurant web queries within three weeks; Saks uses agents for personalized recommendations. Agentforce appeared in roughly 29,000 deals in its first fifteen months.