Workflow · task decomposition · human checkpoints · audit

Charles Trinh — Workflow Automation

A service card reading WORKFLOW AUTOMATION — steps, checkpoints and a trace through every hop
Charles Trinh — Workflow Automation

Automating clinical and administrative workflows without breaking accountability: task decomposition, human checkpoints, and the audit trail through every hop.

What this is

Automation projects usually fail at the seams. Each system does its part correctly and the handoff between them loses the state, the owner and occasionally the patient.

The cure is unglamorous: decompose to steps, give every step a state and an owner, and design the exception path before the happy path.

Done that way, automation is reversible per step and measurable per step, which is what makes it safe to run in a care setting.

Decompose before automating

Automate steps, not departments; a workflow that spans four systems fails at the seams.

Map the steps first and mark the handoffs, because that is where work is lost today. An automation that preserves the seam without fixing it just makes the loss faster.

Human checkpoints

Place a person where being wrong is irreversible, and give them the context to decide rather than a queue.

A checkpoint with no decision content is a click-through. Show the evidence, the alternatives and the consequence, and measure how often the person overrides the automation.

State and ownership

Every task has a state and an owner; automation without state visibility creates work that disappears.

The state machine should be legible to an operator without reading code: named states, allowed transitions, and a view of everything currently stuck.

Exception paths

The exception is the work; design the exception path first and the happy path becomes trivial.

Enumerate the realistic exceptions with the operations team, then build the queue before the automation. Otherwise the first exception parks a task forever.

Audit through hops

Record which system and which person touched each hop, or the trace cannot be reconstructed.

One trace id carried through every hop turns a reconstruction exercise into a query. Add it at the start; retrofitting it later means rebuilding the pipeline.

Change management

Automation moves work between people; say where it went, or adoption fails quietly.

Show each affected role what changed for them, and keep the manual path available while the new one earns trust. Silent removal of a familiar step reads as a defect.

Measurement

Cycle time and rework rate, per step, not per workflow; aggregate numbers hide the bottleneck.

Per-step measurement shows exactly which automation helped and which just moved the delay. Rework rate is the honest companion to speed.

Reversibility

Ship automation that can be turned off per step, because the first version will be wrong somewhere.

A per-step switch means a bad rule is contained to one step and one afternoon. A monolith has to be switched off entirely, which nobody will agree to.

How I do it

The first artifact is a step map drawn with the people who do the work, with every handoff marked. That map usually shows where work is lost today, and it is the baseline every later measurement is compared against.

Automation is then applied a step at a time, each with a state, an owner, a trace id and a switch. The exception queue is built before the automation that feeds it, because the first exception arrives sooner than the first quiet week.

We measure per step from the first release: cycle time and rework rate, reported where the team can see them. A step that does not improve both is reverted, which is only possible because the switch is per step.

Artifacts you keep

  • A step map with the handoffs marked and the current loss quantified
  • A state machine per step with a named owner and a stuck-work view
  • An exception queue built before the automation that feeds it
  • A trace id carried through every hop with a queryable record
  • A per-step kill switch and a cycle-time report by step

Automate the decision or the paperwork

The valuable automation removes a step a person performs repeatedly without judgement. Automating a step that requires judgement produces a system that is fast and wrong, and the wrongness is invisible because the step completed.

Map the workflow and mark each step as mechanical or judgemental. Automate the first category, and for the second, present the evidence and keep the decision with a person.

Exceptions are the design

The happy path is easy; the exception set is the work. A workflow that handles eighty percent and drops twenty into a queue nobody staffed has moved the work rather than reduced it.

Enumerate the exceptions before building, and design the queue: who works it, how it is prioritised and how its age is measured.

Change and versioning

A workflow is code, and it needs the same release discipline: versioned definitions, a test set, and a rollback. Workflows edited in a low-code tool at runtime are the hardest to reconstruct after an incident.

Keep the workflow definition in version control even when the tool has its own editor. The definition is the evidence of what ran.

Measuring the automation

Count the steps removed, the exception rate and the rework rate. A workflow whose exception rate rises is a workflow that has drifted from the process it modelled, and the drift is the finding.

Compare against the pre-automation baseline, which means capturing it before the build. Most automation programmes cannot answer whether they helped because nobody wrote down the starting point.

W

here humans stay

T

he steps where a person must remain are the ones with a clinical consequence or an irreversible external effect. Automate the preparation, present the evidence, and record the decision with the identity of the person who made it.

T

hat record is what makes an automated workflow defensible, and it is also what allows the workflow to be improved, because it shows where the human disagreed.

The rollback

Every automated workflow needs a way to stop and a way to finish what is in flight. Stopping is easy; finishing is the design work, and it is what prevents an outage from becoming a data gap.

Document the stop procedure and test it once, because the first time it is used will be under pressure.

Keep the pre-automation baseline in the repository with the workflow. It is the only way to answer, a year later, whether the automation reduced work or moved it into a queue nobody counts.

Name the owner of each workflow in the definition, not in a spreadsheet. A workflow whose owner is recorded somewhere else is a workflow that is orphaned the first time the spreadsheet is not consulted.

Record the exception rate before and after each change to the workflow, so the effect of the change is measured rather than assumed.

Review the workflow against the manual process annually; processes change and the automation does not unless somebody compares the two on purpose.

Where the workflow touches an external system, record its failure behaviour explicitly: retry, queue or stop. An undocumented choice is discovered during an outage.

Keep the test set of exceptions in the repository and run it on every change, since exceptions are the part of the workflow that most changes quietly break.

The set grows as incidents are reviewed, and it becomes the record of what the workflow has learned.

Proof

  • ONC (g)(10) self-test PASS, 317 of 317 subtests, April 2026 — public suite URL
  • 67 public hostnames, 51 serving, one AWS account
  • 104 clearinghouse enrolments across 53 jurisdictions
  • USCDI v7 proposal — Patient Digital Identity Credential — listed at isp.healthit.gov
  • SAM.gov UEI MVKQW9DKKGB8 · CAGE 20KM5 · NPI 1194645564
  • AWS Business Associate Addendum active since 18 August 2026

Where this work is delivered

Engagements run from Las Vegas and reach across the valley: clients in Las Vegas, NV; Las Vegas, Nevada; Henderson; Clark County, Nevada; and North Las Vegas, Nevada, plus remote delivery nationwide.

How an engagement runs

1. Trace one real case

Follow a single patient or claim end to end with a timestamp at every hop, before any document is written.

2. Written findings

A findings list naming the defect, the system, the owner and the date.

3. Monthly review

Review the list monthly and check whether the dates held.

Frequently asked

What is this service?

This page describes charles trinh — workflow automation — the scope, the failure modes it prevents and how the work is measured. It sits in the AI and Automation category at trinh.io.

Where is the work delivered?

Engagements run from Las Vegas, Nevada across the valley — Las Vegas, Henderson, North Las Vegas, Spring Valley, Paradise, Winchester, Sunrise Manor, Enterprise, Summerlin, Clark County — and remotely anywhere in the United States. On-site work in the Las Vegas valley, remote delivery elsewhere.

What does it cost?

Standard rate $195 per hour, committed $165 per hour at twenty hours or more per week for three months or more; $1,600 standard day, $1,350 committed day. Scoped in writing before work starts, fixed fee where the deliverable is defined.

Who is Charles Trinh?

A technology architect in Las Vegas, Nevada, working in healthcare identity, consent and interoperability. Not a physician, not a medical practice, and not affiliated with any other Charles Trinh.

How does an engagement start?

With one trace of a real case and a written findings list naming the defect, the system, the owner and the date. The first call is twenty minutes and free: 702-559-3489.

Engagement

Scoped in writing before work starts; fixed fee where the deliverable is defined. First call is 20 minutes and free. 702-559-3489 · charles@trinh.io

Talk to Charles

TRINH — Charles Trinh, technology architect
TRINH

Charles Trinh

Fractional CTO/CIO and Technology Architect — Healthcare Data, Identity & Regulated Infrastructure · Las Vegas, NV

Las Vegas, Nevada · US citizen · Fractional and project engagements

(702) 559-3489 Email Charles Trinh