Critical steps were still physical or manually driven
Case Study · 03
How a manufacturing giant replaced physical process handling with digital workflows and practical automation
Critical workflows still ran on paper, spreadsheets and follow-up, with people as the glue between steps. Senatio modeled the process explicitly, digitized it, and automated the repeatable parts - turning effort into a system.
01 · Situation
Process state lived in people
Approvals lived across documents, spreadsheets, calls and memory. The process had states, but nothing recorded them - every question about where work stood meant interrupting a person, and managers ran the operation as human workflow engines.
This burden hides in plain sight: the company looks operational, but the movement is powered by effort instead of design.
How it showed up on the floor
Approvals needed repeated human follow-up
Weak visibility into what moved and where delay was forming
Teams spent their time checking status and correcting drift
Output depended on experienced people pushing the process
Volume was scaling ahead of operating control
02 · Architecture
Give the process an explicit model
The instinct is to build an app that replaces the paper. That digitizes the same problem: forms that capture the end of a step, and a second place for status to disagree with the first.
The decision that mattered: model the workflow as explicit states and permitted transitions. Every transition recorded as an event - who, when, on what basis - with current state a projection of that history. The audit trail exists by construction.
The workflow layer owns process state only. Planning and record systems keep their master data; integration happens at the edge. That keeps the blast radius small enough for a live operation.
03 · Constraints
What the design had to hold
The operation could not stop
No cutover window. Everything runs alongside, reversible on a shift's notice.
The floor is not an office
Shared devices, gloves, noise, broken connectivity. Design for that or get worked around.
Systems of record stay authoritative
The workflow layer must not become a second source of truth.
Judgment steps stay human
Some approvals encode accountability, not logic. Only deterministic transitions automate.
The old path keeps working
Paper and digital describe one reality. The system reconciles them, not operators.
Traceability is not optional
Who moved what, when, on what basis - built into the data model, not bolted on.
04 · Intervention
What Senatio built
First, map the process as it actually ran - rarely the process as documented. States were named in the floor’s own vocabulary, because a system that disagrees with its operators gets worked around.
Capture moved to the point of work. A step recorded where it happens produces data that is true; one transcribed later produces data that is approximately true - worse than none, because it will be trusted.
The automation was deliberately unglamorous: route on state change, raise what sits too long, delete the confirmation step that only ever confirmed.
05 · Production
Shipping into a running operation
The digital flow ran alongside the old one before it ran instead of it, with reconciliation as the system’s job. Rollout went area by area, and every area kept a path back that needed nobody’s permission.
Senatio stayed through stabilization. A workflow system is done when the operation stops using what it replaced.
06 · Measured result
The process became measurable
The chasing stopped. More durable: the operation stopped estimating its own behavior. These became queries, not opinions:
Cycle time per state, not for the process as a whole
Queue depth at each step - where bottlenecks announce themselves
Work that stopped moving, distinct from work that is slow
Touch count - what a unit of work costs in human interventions
Rework and reversal rates, recorded rather than overwritten
Throughput by area and shift, on states that mean the same thing everywhere
Bring us a problem
Tell us what is actually happening
If there is technical work that needs an owner rather than an opinion, that is the conversation worth having.