Salesforce Launches AIforce and Agentforce Coworker: AI Agents Become Enterprise Teammates

At Dreamforce 2026 in San Francisco, Salesforce drew a line under the chatbot era. The company launched AIforce, an interface layer that exposes its entire platform — data, workflows, business logic, permissions and governance — to any external AI agent, and Agentforce Coworker, an AI teammate that operates inside the Salesforce Lightning interface alongside human users. The announcements were accompanied by a concrete deployment commitment: Adecco Group, the world's largest staffing firm, will deploy Agentforce Coworker across 40 countries, enabling recruiters to trigger pre-screening, recruiting and onboarding agents directly from their existing systems.

The move matters not because Salesforce has added another AI feature. It matters because the combination of a composable agent platform, a global enterprise deployment, and a growing portfolio of job-ready agents signals that AI coworkers are moving from internal pilots into core business operations at a scale that will be difficult to ignore. For enterprise buyers and suppliers operating anywhere in the agentic AI ecosystem, the Dreamforce announcements offer a preview of how the market's largest CRM platform intends to make AI agents a default part of the enterprise technology stack.

What changed: three layers, one architecture

Salesforce's announcements at Dreamforce rest on a three-layer architecture that the company calls its Agentic Enterprise stack. Data 360 provides harmonised and federated data, metadata and memory so every agent understands the customer and the business. Customer 360 supplies the business logic, processes, permissions and actions that agents need to operate across sales, service, marketing and commerce. And now AIforce sits on top of both, exposing that entire foundation to any AI interface — Claude, Slack, Microsoft Teams, Agentforce agents, or custom-built interfaces — while enforcing the same permissions and governance that already apply inside Salesforce.

The practical effect is significant: employees who have never opened a Salesforce dashboard can ask questions, update records and trigger workflows from whatever tool they already use, and every action routes back through Salesforce's existing security model. There is no new permissions system to configure, no migration project to fund, no custom integration work to maintain. The company says 100,000 users activated Agentforce Coworker within its first 35 days.

Alongside the architectural announcement, Salesforce introduced seven job-ready agents with defined roles: Casey for customer service, Paige for IT and HR, Carter for shopping and commerce, Hunter for outbound sales, Marshall for supply chain, Piper for inbound pipeline generation, and Fin for complex customer experience workflows. Six are generally available now; Hunter remains in pilot with general availability planned for November 2026 and is the first agent to run on a new long-horizon runtime that lets agents pursue goals over days and weeks rather than within a single chat session.

Adecco across 40 countries: the proof point

The headline deployment commitment came from Adecco Group. The company, which operates in 60 countries and places hundreds of thousands of workers annually, will deploy Agentforce Coworker across 40 countries. The stated goal is to let recruiters trigger pre-screening, recruiting and onboarding agents from their existing systems, embedding agentic workflows directly into the core business process rather than running them alongside it.

This is not a pilot. A deployment across 40 countries in a workforce of Adecco's scale represents one of the largest known enterprise commitments to agentic AI in a people-intensive, compliance-heavy industry. Staffing and recruitment involve sensitive personal data, employment law obligations that vary by jurisdiction, and workflows that require human judgement at multiple stages. If Agentforce Coworker can operate safely within those constraints at Adecco's global scale, it establishes a precedent that will be cited by enterprise buyers in every regulated sector.

Adecco was not the only customer named. Siemens has deployed Agentforce to engage 100 per cent of its inbound leads. Fulton Bank reported moving from zero to more than 20 use cases live in production within weeks, supporting approximately 3,000 users. Engine, a travel insurance provider handling over 800,000 customer inquiries annually, credited Slackforce with enabling anyone on the team to describe what they need and receive a real, working interface built from data the company already holds. Across the announced customer base, Salesforce reported delivering 7 billion Agentic Work Units since the platform launched, including 3.2 billion in the second quarter of 2026 alone.

What the metrics tell us

The customer results Salesforce shared are worth examining because they provide benchmarks for what production AI agents can achieve today, rather than what they might achieve in the future:

• 50 per cent of Engine's chat inquiries are fully resolved by its help agent
• 60 per cent of Perk's sales pipeline is built by its outbound sales agent, Hunter
• 70 per cent of Autism Queensland's administrative requests are resolved by its employee service agent, Paige
• 90 per cent of core shopper journeys for Hibbett AI went live in six weeks
• Four times the conversation volume is driven by Asana's website agent, Piper, with deployment averaging 45 days
• 79 per cent of Anthropic's conversations seen by Fin are resolved autonomously

These are not vanity metrics. A help agent resolving 50 per cent of chat inquiries and a sales agent building 60 per cent of pipeline are delivering measurable operational leverage. The average deployment time of 45 days for Asana's Piper agent suggests that the gap between deciding to deploy an agent and seeing results is shrinking to weeks rather than quarters. For enterprise buyers building business cases for agentic AI in the second half of 2026, these numbers provide a credible reference point.

The long-horizon runtime: agents that don't clock out

One of the most significant technical announcements is the long-horizon runtime, which lets agents pursue goals over days and weeks rather than completing only a single task or interaction. Hunter, the outbound sales agent, is the first to use it. A seller might ask Hunter to rescue at-risk deals before the end of the quarter; Hunter turns that into a measurable goal, builds a plan, and begins working toward it across multiple sessions, incorporating new information and adjusting the plan while keeping the seller in control.

The runtime is built on three technical capabilities. Memory preserves context and progress across sessions so work does not stop when the interaction ends. Durable execution keeps plans running and allows an agent to resume or course-correct as circumstances change. Dynamic steering adapts an agent's behaviour based on individual user feedback and direction. Together, they represent a meaningful step beyond the single-interaction model that has defined most enterprise AI deployments to date.

The governance layer arrives at the same time

The Dreamforce announcements arrived in the same week that WSO2 released Agent Manager, an open control plane for governing AI agents across any framework, model or deployment, under an Apache 2.0 open-source licence. WSO2's timing is instructive. The company cited Gartner's prediction that the average Fortune 500 enterprise will have more than 150,000 agents in use by 2028, and noted that only 13 per cent of organisations believe they have adequate AI agent governance in place.

Salesforce addressed this directly with a Trusted Enterprise AI Harness and an AI Control Plane that registers agents, sets identity and policy, manages lifecycle, evaluates performance, observes behaviour and controls cost across both Salesforce and third-party AI. The simultaneous emergence of proprietary and open-source governance layers illustrates where the market is heading: deployment and governance are no longer separate conversations. Enterprises are demanding both at the same time, and vendors that deliver integrated governance will capture more trust, and more budget, than those that treat it as a later add-on.

What it means for enterprise buyers

The Dreamforce announcements carry five practical implications for organisations evaluating agentic AI platforms.

Pre-built agents reduce the build-versus-buy equation. Salesforce's seven job-ready agents cover sales, service, commerce, HR, IT and supply chain. For organisations already running on Salesforce, the time from procurement to production deployment is now measured in weeks, not months. For organisations on other platforms, the announcement raises the competitive bar: if your CRM or ERP vendor is not offering pre-built agent roles with defined skills and actions, you are paying a time-to-value tax that your competitors are not.

Agentic AI is arriving through existing platforms, not alongside them. The AIforce architecture is significant precisely because it does not require a separate deployment. It exposes what already exists inside Salesforce through new interfaces. For procurement teams, this means agentic AI procurement increasingly looks like platform procurement. The decision about which agents to deploy is inseparable from the decision about which platform to run them on.

Long-horizon agents change the ROI conversation. An agent that pursues a goal over weeks — rescuing at-risk deals, managing a hiring pipeline, monitoring supply chain disruptions — creates value that a single-interaction chatbot cannot. But it also creates new operational risks around agent autonomy, error propagation and compliance. Buyers should evaluate long-horizon capabilities alongside governance features, not separately from them.

Deployment velocity is becoming a competitive metric. Asana deployed Piper in 45 days. Hibbett AI went live in six weeks. Fulton Bank launched 20 use cases in weeks. The time horizon for agentic AI projects is compressing, and organisations that treat agent deployment as a multi-quarter transformation programme will find themselves outpaced by competitors who treat it as a series of weeks-long sprints.

Governance is now a platform feature, not a consulting engagement. Both Salesforce's AI Control Plane and WSO2's open-source Agent Manager are designed to be turned on, not built from scratch. For buyers, this shifts the governance conversation from "how do we build it" to "which platform's governance model fits our risk profile and regulatory obligations."

What it means for suppliers

For companies building or selling agentic AI platforms, three dynamics stand out from the Salesforce announcements.

The platform giants are raising the integration bar. AIforce exposes Salesforce's entire data model, permissions framework and business logic through MCPs, APIs, plug-ins and skills. For independent agent builders, the question is no longer whether they can build a capable agent. It is whether their agent can plug into the platforms where enterprise data already lives with the same depth of integration that Salesforce's own agents offer. Deep platform integration is becoming table stakes.

Open ecosystems create both opportunity and pressure. Salesforce's Headless Toolkit and AgentExchange marketplace create a distribution channel for third-party agent builders to reach the Salesforce installed base. But they also mean that independent agents compete directly with Salesforce's own job-ready agents. Suppliers that differentiate on vertical expertise, domain-specific skills or multi-platform orchestration will fare better than those offering generic capabilities that platform vendors can ship themselves.

The governance market is forming alongside the deployment market. WSO2's Agent Manager launch in the same week as Dreamforce is not a coincidence. As platforms like Salesforce, Microsoft, Google and AWS make agents easier to deploy, the demand for cross-platform governance, observability and control grows in parallel. Independent governance vendors have a window to establish themselves before the platform vendors close the gap.

The Agentic Expo angle

Dreamforce 2026 answered a question that has hung over the agentic AI market all year: when do AI agents stop being a technology story and become an operations story? The answer, judging by Adecco's 40-country deployment and the seven job-ready agents now available to the Salesforce install base, is now.

Agentic Expo 2027, taking place at Olympia London on 23-24 March 2027, sits at the intersection of the trends these announcements illustrate. The exhibitors and speakers will represent every layer of the stack that Salesforce, WSO2 and their competitors are building: the platforms, the governance tools, the professional services firms and the industry specialists who turn platform capabilities into production outcomes. A market where 100,000 users activate an AI coworker in 35 days is a market that needs a dedicated space for buyers and suppliers to evaluate what works, what is secure and what is worth betting on.