Salesforce has reported that its "agentic AI workforce" — autonomous AI agents deployed by its customers to handle business tasks — is more than doubling year on year, according to Salesforce News. The figures come from Salesforce's own platform data and reflect usage across its Agentforce product line and related agentic tooling.
The report frames this growth as evidence that agentic AI — systems that can execute multi-step tasks with a degree of autonomy, rather than simply generating text or answering single queries — is moving out of the pilot phase and into standard operational use across the companies Salesforce serves. Salesforce has not, as far as this report discloses, broken down the growth figures by company size, industry, or specific use case, so it remains unconfirmed how much of this expansion is concentrated among large enterprises with dedicated implementation teams versus smaller companies adopting off-the-shelf agent products.
For a 10-200 person B2B company, the number itself is less important than what it implies about the state of the underlying technology. A vendor citing sustained, more-than-doubling growth in deployed agents is effectively saying that enough customers have gotten past the setup, integration, and trust-building hurdles that agentic AI is no longer confined to flagship pilot programs. That matters for smaller operators specifically because they typically lack the internal resources to build custom agent infrastructure from scratch — they depend on the ecosystem around platforms like Salesforce (or comparable tools) reaching a level of maturity where a narrow, well-defined workflow can be automated without months of custom engineering.
The most direct relevance for smaller B2B teams sits in three areas: sales lead qualification and routing, first-line customer support triage, and repetitive back-office operations such as data entry between systems or order status updates. These are the categories where agentic tools are most commonly deployed today, largely because they involve clear inputs, bounded decision spaces, and measurable outcomes — properties that make them tractable for current-generation agents in a way that more judgment-heavy work is not yet.
Two cautions are worth stating plainly. First, Salesforce's growth figures describe usage of its own ecosystem and should not be read as an independent, third-party benchmark of the broader agentic AI market; other vendors and independent research firms have published different adoption estimates, and this report does not reconcile them. Second, "doubling" says nothing about deployment quality, agent reliability, or return on investment — a metric a company evaluating agentic AI tools should demand from any vendor before committing budget, rather than inferring it from aggregate growth statistics alone.
The practical signal for operators is that agentic AI adoption is no longer a leading-edge bet confined to large enterprises. Whether that translates into value for a specific 10-200 person company still depends entirely on picking a narrow, well-scoped process to automate first, rather than treating "agentic AI" as a single monolithic capability to switch on.