Salesforce Unveils Agentforce Operations to Tackle Workflow Breakdowns in Enterprise AI

Salesforce has launched Agentforce Operations, a new workflow platform designed to fix the broken processes that are crippling enterprise AI deployments. The platform imposes a deterministic structure on workflows that agents are expected to execute, addressing a critical failure point where tasks fail and handoffs break down.

According to Sanjna Parulekar, Salesforce senior vice president of Product, many enterprise workflows were never built for agents. “What we’ve observed with customers is that a lot of times, the brokenness in a process is probably in your product requirements document,” Parulekar told us in an interview. “So when that’s uploaded into a product, it doesn’t quite work. We can optimize it and cut out some things and replace it with an agent.”

Without this control panel layer, enterprises risk deploying agents that increase costs rather than fix workflow problems. The platform turns back-office workflows into a set of tasks for specialized agents, allowing users to upload their processes or use Salesforce-provided Blueprints.

The Problem: Workflows Designed for Humans, Not Machines

Enterprises deploying agents are learning a costly lesson: their workflows were designed around human judgment gaps, not machine execution. Processes that evolved through years of workarounds — loosely defined steps, implicit decisions, coordination that depends on individuals knowing what to do next — break when agents are asked to follow them literally.

Even with all of an enterprise’s context at its fingertips, AI systems struggle to complete tasks if it’s not clear what they’re supposed to do. Parulekar noted that focusing on what makes a process tick and breaking it down into more explicit steps makes the system more deterministic. Then, when platforms like Agentforce Operations introduce agents, those agents already know their specific tasks.

“It forces companies to rethink their processes and introduces observability into the mix because of the session tracing model in the system,” she said. Human checks can be built into the system, making the process more transparent.

What Makes This Different

Unlike traditional workflow automation tools that rely on agents to decide what to do next based on probabilistic decision-making, Agentforce Operations enforces execution on a pre-defined, deterministic structure. This means the system controls the workflow, not the agent, reducing ambiguity and failure rates.

The Catch: Codifying Flaws at Scale

However, the approach introduces a new risk: if a process has flawed steps, encoding it for agents locks in the problem at scale. “Codifying a workflow doesn’t fix a broken one,” Parulekar cautioned. Enterprises must therefore audit their processes before adopting this layer.

Background

Enterprise AI teams are hitting a wall, not because their models can’t reason, but because the workflows underneath them were never built for agents. Tasks fail, handoffs break, and the problem compounds as organizations push agents deeper into back-office systems. A new architectural layer is emerging to address it: workflow execution control planes that impose deterministic structure on processes agents are expected to run. Salesforce is one of the companies bringing this to the forefront.

What This Means

For enterprises, Agentforce Operations represents a shift from hoping agents will improvise to controlling their execution with precision. It promises to reduce agent failures and increase transparency, but it also demands a rigorous reexamination of existing workflows. Companies that invest in process auditing and redesign will benefit most. Those that skip this step may find themselves scaling inefficiency faster than ever.

As AI agents move from chat interfaces into core business operations, the ability to manage workflow execution will become a critical competitive advantage. Salesforce’s move signals that the industry is recognizing that the real bottleneck isn’t AI intelligence—it’s the broken processes it is asked to run.