Orkes Case Study | Stepwork

A durable orchestration layer helped Stepwork reduce scheduling overhead, diagnose failures faster, and pursue enterprise-scale automation with confidence.

Automation Beyond the API Boundary

Stepwork is a deterministic interface automation platform that runs repeatable workflows across applications, including systems without APIs. Unlike purely LLM-driven agents, it uses rule-based execution for predictable outcomes and applies AI only where reasoning is useful. Its flows handle everything from employee onboarding and offboarding to IT access reviews and vendor data extraction, work that’s too inconsistent for pure API automation.

Before Orkes, Stepwork ran on a homegrown scheduler built to handle the “happy path,” the simplest, most predictable version of a workflow. As customer flows grew more complex, so did the engineering cost of maintaining it. Stepwork made the call to buy rather than keep building, and adopted Orkes Conductor to power scheduling, orchestration, and auditability across every flow it runs.

The Challenge: A Homegrown Scheduler That Couldn’t Keep Up

Stepwork’s product depends on flows that chain together, sometimes 10 to 20 steps deep, passing data from one system to the next: pulling a new hire’s details from an IT helpdesk, then feeding that data into Google Workspace or Microsoft 365, for example. Each flow needs to run reliably, on schedule, and in the right sequence.

Early on, Stepwork built its own scheduling system on top of basic AWS primitives. It worked for simple cases. But every new feature, like flows within flows, org-level batching, or user-specific scheduling, meant more engineering time spent maintaining infrastructure instead of solving customer problems. Over time, orchestration became a recurring tax on the engineering team.

“The switching cost of building and maintaining our own scheduler was the real hidden cost. We bought Orkes, it works. We’re not rebuilding this every day for some edge case a customer comes up with.”

— Shaun MacLellan, CEO, Stepwork

Stepwork estimates it spent roughly two months of combined time from a senior and a junior engineer just getting an early version of the in-house scheduler production-ready, time MacLellan felt was better spent on customer-facing problems. The larger cost, however, was not the initial build. It was the repeated context switching required whenever a new scheduling edge case appeared.

Stepwork made a deliberate architectural choice: scheduling and workflow orchestration needed to be reliable infrastructure, not an internal product line. The team selected Orkes Conductor so engineers could focus on the automation capabilities customers buy from Stepwork rather than rebuilding workflow orchestration. Stepwork evaluated a small set of orchestration options, including Temporal, before choosing Orkes.

How Stepwork Uses Orkes

Orkes now sits underneath nearly every flow Stepwork runs, across a range of scenarios:

Scheduled, credential-based automation

Stepwork can automatically sign into customer applications and schedule flows to run unattended, including overnight, without a human in the loop to approve access. This requires a scheduling engine reliable enough that a flow triggered at 2 a.m. works every time.
Distributed execution across customer environments

Flows run across customer clouds and local devices, with specific users or devices targeted per flow. Orkes handles the orchestration layer while execution happens on the customer’s own infrastructure.
Org-level and individual-level scheduling

Stepwork schedules large batches at the organization level and at the individual user level. The same batch scheduling lets Stepwork pull multiple workers at once for internal testing, validating flows at scale before they reach a customer’s production environment.
Environment separation

Separate development and production environments give Stepwork’s team confidence to test aggressively without risking production traffic.

Making Complex Workflows Explainable

When a selector changes or a website behaves differently mid-flow, operators need to know which of those 10 to 20 chained steps failed, and what data entered and left that stage.

“If a customer offboards a rogue employee heading to a competitor, they don’t want that fixed next week. They want it fixed today,” MacLellan said. “Having a full audit trail means we can diagnose and fix issues the same day, instead of going on a scavenger hunt through logs.”

Conductor’s visual workflow model and execution history give Stepwork that line of sight. The team can inspect the path through a procedure, audit inputs and outputs, and isolate a failure without searching through the raw logs of every microflow. That visibility matters for time-sensitive workflows. Stepwork has even mirrored the workflow structure inside its own diagnostic experience, allowing teams to cross-reference customer-facing procedures with the underlying orchestration data.

“Even as a non-technical person, I can look at the data and get a general sense of where something failed.”

— Shaun MacLellan, CEO, Stepwork

Results

Scale, with confidence

Stepwork now runs 100 to 200 complex flow executions per day, and works with enterprise customers running up to 500 complex flows every 30 days across core vendor processes. MacLellan said Stepwork wouldn’t have been comfortable offering this volume to customers without the audit trail Orkes provides.

A stronger enterprise sales story

Being able to point to Orkes as a proven, auditable scheduling layer has made it easier for Stepwork to close pilots and contracts with large enterprise customers who scrutinize every subprocessor.

Business model alignment

Because Stepwork charges on a consumption basis, every flow that runs reliably and without manual intervention directly supports revenue. Fewer approvals and less human-in-the-loop friction mean more consistent volume for both companies.

Engineering time redirected

Day-to-day, Stepwork’s engineering team spends very little time managing Conductor itself. What time is invested goes toward extending orchestration for new use cases, like multi-device and multi-cloud execution, rather than maintaining core scheduling infrastructure.

A Durable Layer Beneath Differentiated Automation

For Stepwork, the build-versus-buy question became straightforward. The company’s differentiation lies in turning user-interface interactions into reliable automation across systems that APIs cannot fully reach. Orkes Conductor supplies the scheduling, coordination, and visibility underneath that experience.

With that foundation in place, Stepwork can direct engineering effort toward the problems customers notice: broader automation coverage, dependable execution, rapid diagnosis, and less manual work in the loop.