Salesforce has introduced Agentforce Operations, a solution that turns obsolete and manual back office processes into a set of tasks where specialized agents perform process coordination, data verification, regulatory compliance, and seeking approvals, among others. The result is a 50% to 70% reduction in cycle times and an 80% reduction in manual data entry.
Every day, companies use more agents to make experiences with their customers and employees faster and more personalized. When these modern, fast experiences are confronted with a web of outdated, manual back office processes, everything stops. Employees are caught switching between systems, pasting data into spreadsheets, and checking email for updates. What should take hours can take days or weeks; and means a slowdown in productivity, a delay in income and an increase in costs.
“As companies accelerate the adoption of AI to become Agentic Companies, most remain burdened by an underlying layer of fragmented and manual processes in the supply chain, purchasing, finance and the general back office. This silently slows down operations, increases costs and limits growth. With Agentforce Operations, we are not only digitizing those processes, but rethinking them for an AI-centric environment, optimizing how agents and humans work together,” explained Aman Naimat, SVP & GM of Agentforce Operations.
End-to-end agentic execution
Agentforce Operations extends the capabilities of AI agents from front office experiences into the processes and systems that run the business, enabling end-to-end agentic execution in the back office. This is what it looks like in different industries:
- For any manufacturer, with Agentforce Operations, agents orchestrate the end-to-end fulfillment process: verifying inventory, coordinating across teams and suppliers, managing approvals across systems, and proactively activating Field Service workflows to schedule on-site installation, delivering product on time, and giving the salesperson a head start on the next deal.
- For a bank, a customer submits a loan application and expects a quick decision. With Agentforce Operations, agents manage the entire loan underwriting process—extracting data from tax returns, chasing down missing signatures, and validating every detail against compliance standards across multiple systems—so loan officers can focus on the customer instead of the paperwork.
- For an insurer, a customer files a claim and expects a quick and fair resolution. With Agentforce Operations, agents coordinate the claims receipt and validation process—verifying details, tracking missing information, and assembling a complete file—accelerating claims management and accuracy to avoid subsequent delays.
Structured and automated processes
Workflow automation has been around for years. But most legacy platforms were designed to coordinate steps, not complete complex jobs from start to finish. While these solutions can route tasks, manage approvals, and improve visibility, when work relies on unstructured information, or spans multiple systems and is constantly changing, teams still rely on people and technical specialists to remain operational. The result is brittle automation that breaks down at boundaries between systems, bottlenecks that slow down processes, and an inability to scale.
Agentforce Operations introduces a new model for back office execution. AI agents do more than just orchestrate work. They complete it, adapt as the business changes, and coordinate with people only when necessary. End-to-end processes run faster, require less manual intervention, and remain accurate and auditable as they evolve.
1.- Smart operations: Agentforce Operations coordinates tasks, deadlines and logistics between AI agents that execute work autonomously and humans when necessary:
Specialized agents autonomously complete repetitive tasks such as extracting data from complex documents, running calculations, updating credit models, and identifying compliance gaps. What once took a team of auditors four hours, now takes an AI agent minutes, delivering consistent, high-quality results with a complete audit history.
Frictionless adoption: People continue working within their usual tools while AI agents keep processes running in the background. Whether you’re a customer, employee, broker, or supplier, you can interact with tasks directly from email, as well as from Slack and Microsoft Teams.
Accurate results at scale: Critical processes need to be correct at all times. By combining LLM model reasoning with business rules, agents interpret complex or disorganized information and complete it at every step exactly as needed, delivering accurate, large-scale audit-ready results.
2.- Instant plans: Agents and humans need a single set of instructions to work together. Agentforce Operations converts unstructured documents or diagrams into “digital blueprints” in minutes.
3.- Adaptive, transparent and in continuous improvement: Most automated processes fail because modifying them requires a large number of developers. With Agentforce Operations, business leaders can make their own changes:
Natural language updates: If a new regulation appears, a manager can update the entire operation simply by describing the change in an email, and it is updated.
“Anti-bottleneck” engine: Agents proactively detect delays—such as a three-day lag in signatures—and suggest solutions before they affect the customer.
Radical transparency: Every AI action is recorded and linked digitally, providing a permanent audit trail. Instead of having to review files months later to find an error, an active security network keeps your business organized and prepared for audits.
Agentforce Operations is built on Salesforce’s Regrello technology, which has transformed the world’s most demanding supply chains, reducing time, eliminating manual coordination, and improving visibility into complex operations.
For availability purposes, Agentforce Operations is already available for use. Ecosystem integration capabilities – such as the ability to automatically sync data and trigger actions with Salesforce Flows – are expected to enter beta in May 2026.
