Internal Tools

Manual vs Automated Process: Choose an Operating Model, Not a Side

For an owner deciding how much of a process should run in software, which parts should stay human, and how the exception path will work on a bad day.

Uli PrantzBuilds and operates all-agents
Published
Key takeaways

Key takeaways

  • Manual, assisted, and automated are operating models, not maturity levels.
  • Manual work adapts cheaply to novelty but scales poorly and often hides state in personal queues.
  • Automation improves consistency and visibility only when the process, controls, and evidence are designed with it.
  • A hybrid process is often strongest: code moves the normal path while people own consequential judgment and exceptions.
  • Compare correction cost and failure impact alongside time saved.
A manual process is flexible at the point of work but depends on human attention, memory, and capacity. An automated process is consistent and observable when its rules, state, exceptions, and controls are defined, but it introduces software ownership and can repeat mistakes at speed. An assisted process often gives the strongest first operating model: software assembles, checks, drafts, and records while a person handles novelty or approves consequential effects.

Compare three operating models, not two

No column is universally more mature.
DimensionManualAssistedAutomated
Novel casesAdapts immediately if the operator understands themSoftware prepares; person resolvesMust route outside the bounded path
ConsistencyVaries by person, load, and interpretationStable checks with human judgmentStable inside encoded rules and tested judgment
CapacityScales with trained attentionRemoves preparation and coordinationScales until dependency, exception, or control limits
EvidenceOften scattered unless deliberately recordedSystem assembles; person confirmsCan record every state and effect if designed to
Change costTraining and coordinationInstructions, tests, and interfaceVersioning, tests, release, monitoring, and communication
Failure shapeMissed, inconsistent, or delayed workReview burden or misplaced trustRepeatable failure with potentially broad scope

Choose by variability, consequence, and checkability

Keep work manual when legitimate cases differ so much that rules would be fiction, the policy is still moving, or the outcome cannot be checked promptly. Use assistance when inputs are messy but a person can verify a proposal. Automate a bounded path when inputs, rules, output, and exception destinations are stable enough to test.

The internal tool is the surface people may use in any of these models. A manual process can run through a simple case tracker. A fully automated background step may need no interface. Approvals and exception queues usually do.

The strongest first move can be a partial redesign.
SituationFirst move
Rare, high-novelty reviewDocument principles, owner, and evidence; keep judgment manual
Frequent data gatheringAutomate collection and validation; keep the decision human
Stable reversible updateAutomate with idempotency, verification, and exception routing
Consequential communicationDraft automatically; require accountable approval before sending
Unowned or disputed processResolve ownership and policy before building

Design the exception model before changing the normal path

  1. 01Collect recent exceptionsHuman approval
    Use actual missing, conflicting, duplicate, stale, urgent, and policy-edge cases.
  2. 02Classify the destinationCode
    Reject invalid input, request information, retry a transient step, reconcile an ambiguous effect, or create owned review work.
  3. 03Set a clock and ownerTrigger
    Every exception has a visible state, responsible role, due time, and escalation.
  4. 04Return or closeHuman approval
    State whether the resolved case returns to the normal path or closes with a different outcome and evidence.

Move through observe, assist, and bounded automation

Google's automation guidance describes automation as a force multiplier rather than a panacea. The same point governs transition: prove where the force is pointed before increasing it.

  1. 01BaselineHuman approval
    Measure volume, elapsed time, touches, corrections, exceptions, missed work, and harmful outcomes.
  2. 02ObserveAI judgment
    Generate a parallel recommendation without changing external state; inspect disagreements.
  3. 03AssistHuman approval
    Prepare work automatically while the operator verifies each outcome and records corrections.
  4. 04Automate a safe subsetCode
    Release the stable, checkable, reversible path with strict scope and fallback.
  5. 05ReviewHuman approval
    Compare clean completion, correction, exceptions, detection, recovery, and owner burden before expanding.

Limitations and when not to use this

  • Manual work is not automatically uncontrolled, and automated work is not automatically consistent. Both depend on design, evidence, and ownership.
  • A hybrid process can add review burden if the automated proposal is unreliable or the person cannot see enough evidence to verify it.
  • Consequential and regulated decisions may require permanent human accountability regardless of technical capability.

Sources

  1. The Evolution of Automation at GoogleGoogle Site Reliability Engineering Accessed 14 August 2026
  2. Eliminating ToilGoogle, The Site Reliability Workbook Accessed 14 August 2026

Assess the current process

Score its trigger, finish state, normal path, exceptions, observability, owner, and safety before changing the operating model.

Assess the current process
About the author

Uli Prantz

Builds and operates all-agents

Uli Prantz builds all-agents, the process-automation platform this site documents. He writes about the operational side of automating recurring business work: where deterministic code beats model judgment, where it does not, and where a human still has to approve.

Bring one process. We will scope it in 30 minutes.

You leave the call knowing whether it is a fit, what can become code and what still needs a person.

Book a discovery call