# Automation

Automation is software that starts a repeatable task from a trigger and carries it through defined actions, checks, and exceptions with limited manual handling.

Source: https://rodamora.com/glossary/automation
Updated: 2026-08-09

---

Automation moves repeatable work from a trigger to a checked result. AI is optional. Rules work well when the inputs and branches are known. A model helps when the work contains language, images, or other variation that rules cannot handle cleanly. Across the 150+ franchise units in the delivery data I review, roughly 80% use chatbots, 15% run rigid workflows that fire on triggers, and 5% run autonomous agents defined by conditions. The important questions are who starts the run, who chooses the next step, and what proves it finished. The [International Labour Organization assessed nearly 30,000 tasks](https://www.ilo.org/publications/generative-ai-and-jobs-2025-update) in May 2025 and found that one in four workers held an occupation with some generative-AI exposure. Most jobs still mix automatable steps with judgment that stays with a person.

## What does business automation include?

Business automation covers the full path from the event that starts work to the record that proves it finished. A trigger by itself does not complete a process.

Microsoft describes [cloud flows](https://learn.microsoft.com/en-us/power-automate/overview-cloud) with three common start patterns. An automated flow begins after an event. An instant flow begins when a person selects a button. A scheduled flow begins at a set time. The flow then runs actions chosen in advance.

Take new-client intake. A signed agreement starts the run. Rules create the account, open the project, assign the delivery owner, copy the approved scope, and create the first deadline. The final check confirms that those records point to the same client and agreement.

The workflow also needs an exception path. A duplicate account, missing scope, failed write, or unknown delivery owner should stop in a visible queue. The run record should show what finished, what failed, and who owns recovery.

| Part | Question it answers | Intake example |
| --- | --- | --- |
| Trigger | What starts the work? | The signed agreement enters the approved folder |
| Input | Which record owns the facts? | The signed agreement and approved scope |
| Actions | What happens in what order? | Create the account, project, owner, and deadline |
| Check | What proves completion? | Every record points to the same client and scope |
| Exception | What stops or routes the run? | A duplicate account goes to operations |
| Owner | Who repairs failure? | The process owner receives the failed run |

The table is a useful design check. If the team cannot name an answer for one row, the automation has an open operating question.

## What is the difference between automation and AI automation?

Ordinary automation applies known rules to predictable inputs. AI automation adds a model to a step where language, images, or variable input makes those rules hard to write.

Consider a shared intake mailbox. Rules can save the message, match the sender to a client, create a ticket, and set a due date. A model can read the message, classify the request, and draft a summary when each client uses different wording. Another rule can route the result by category.

In an [AI workflow](https://rodamora.com/glossary/ai-workflow), the model handles an uncertain step while the surrounding path remains fixed. Using a model to copy a known account number from one system to another adds variation to a step that already has a clear rule.

The model does not remove the source, check, or exception owner. It changes the kind of input the workflow can handle. If the message contains a scope change or a promise to a client, the system should send the decision to a person before acting.

## When is a workflow better than an AI agent?

Use a workflow when the path can be drawn before the run begins. Use an [AI agent](https://rodamora.com/glossary/ai-agent) when the next action depends on information discovered during the work.

[Anthropic's December 2024 guidance](https://www.anthropic.com/engineering/building-effective-agents) makes this distinction directly. Workflows follow predefined code paths. Agents choose their process and tool use as they go. Anthropic recommends starting with the simplest design that works because more autonomy adds cost and can compound errors.

| Design | Who chooses the next step? | Good fit | Main operating cost |
| --- | --- | --- | --- |
| Fixed automation | Rules written in advance | Known inputs and branches | Maintaining connections and exceptions |
| AI workflow | Rules control the path; a model handles a bounded step | Variable language inside a stable process | Testing model output and routing uncertainty |
| AI agent | The model chooses among allowed actions | A goal where the next action changes with evidence | Tracing decisions, permissions, and recovery |
| Human decision | A qualified person | Hard-to-check or hard-to-reverse consequences | Attention and response time |

A proposal workflow usually has a known path. It collects the approved transcript, matches the account, extracts requirements, drafts the proposal, sends price review to a person, and files the final version. A support investigation may need an agent because the next step depends on what the first system reveals.

Use that flexibility only when the firm can still test the outcome, stop the run, and recover partial work. A longer path gives more places for a wrong decision to travel.

## What should a service firm automate first?

Start with one repeated handoff that has a clear source of truth and a visible completion record. Intake, document routing, recurring reporting, and status collection are common candidates because they happen often and leave inspectable records.

[Write the current path](https://rodamora.com/blog/document-the-process-first-or-the-agent-scales-the-chaos) before choosing a tool. Name the trigger, approved inputs, actions, reviewer, exception, and final record. Test five cases: a normal run, missing information, a duplicate record, a tool failure after one write, and a consequence that requires approval.

Watch enough real runs to find the exceptions people handle from memory. The person doing the work may know that one client uses two account names or that a signed change order outranks the original scope. If that judgment stays unwritten, the automation will follow the cleaner-looking record and fail in the same way each time.

Do not automate six workflows at once. Finish one operating loop. Watch it in production, repair its exception path, and prove that the final record is correct before opening the next one.

## What breaks after an automation launches?

Automations often fail through partial work. One step succeeds, the next times out, and two systems now disagree about the client.

Connections expire. A field changes its name. An API slows down. Someone moves a spreadsheet column. A retry creates the same record twice. A model receives an unfamiliar document and returns a confident category that the workflow accepts. Each case needs a known state and a recovery action.

Microsoft's [current flow limits](https://learn.microsoft.com/en-us/power-automate/limits-and-config) show why maintenance matters. Under some plans, a flow with continuous errors can be turned off after 14 days, and an inactive flow can be suspended after 90 days. Those rules belong to Microsoft's product. The operating lesson is wider: software that ran yesterday can stop after the surrounding system changes.

NIST's [AI Risk Management Framework Core](https://airc.nist.gov/airmf-resources/airmf/5-sec-core/) treats production monitoring as ongoing work. It calls for oversight roles, production monitoring, and post-deployment plans for override, incident response, recovery, and change management. Those duties need an owner after launch.

## Where should a person stay in the workflow?

Keep a person before an action whose consequence is expensive to check or [hard to undo](https://rodamora.com/blog/delegate-the-inputs-own-the-outputs). That usually includes money, scope, a promise to a client, permission changes, deletion, and work where source records conflict.

Give the reviewer a decision packet instead of a blank approval button. Show the source, proposed action, rule that caused the stop, and what happens after approval. A reviewer who must rebuild the case from five systems is still doing the manual process.

NIST notes in its [human-AI interaction appendix](https://airc.nist.gov/airmf-resources/airmf/appendices/app-c-ai-risk-management-and-human-ai-interaction/) that some systems may not require human oversight. It also says operational uses need clear human roles and responsibilities. The boundary depends on the situation and the consequence.

A manual approval is the point where judgment stays with a person. The workflow needs a clear rule for which decisions reach that gate and who receives them.

## How should a firm measure automation?

Measure the finished business record before measuring time saved. Track completion, defects, review time, exception rate, cycle time, and duplicate or conflicting writes.

Then decide [where the recovered capacity goes](https://rodamora.com/blog/the-extra-time-ai-buys-you-is-already-gone). A task can run in two minutes instead of twenty while the employee spends the difference checking errors or waiting on the next approval. The automation [moved cost into another queue](https://rodamora.com/blog/constraint-whack-a-mole-what-ai-adoption-actually-feels-like).

Checking can improve like any other part of the system. In one week of output across many projects in a franchise network's delivery data, more than 300 generated artifacts showed roughly 95% fewer artifacts marked as containing incorrect information or data after tuning, compared with the same system as first implemented. The checking layer used online fact checks, checks against client data, recurring-error rules, and guardrails for repeated language and logic failures. That result is not a benchmark for another firm or the accuracy of one checker. It shows that review is system work that can be measured and changed.

## Where does automation sit on the Delivery Model Ladder?

Automation belongs mainly at Stage 2, Augmented, on the [Delivery Model Ladder](https://rodamora.com/delivery-model-ladder). The workflow starts from a defined event, carries work across systems, checks the result, and routes exceptions to a named person. The firm has changed how delivery runs.

At Stage 1, Enhanced, a person may push the button, inspect every result, and copy it into the next system. The tool makes that person's task faster while the surrounding workflow stays the same.

At Stage 3, AI-native, larger parts of delivery can start from conditions and run across shared company systems. More work happens without a person assembling every input or handoff. Testing, monitoring, and recovery matter more because one failure can travel farther before a person sees it.

The gap between a working demonstration and an owned operating system belongs to the [Production Gap](https://rodamora.com/production-gap). Automation closes part of it when the trigger, checks, exceptions, and recovery owner travel with the work.

## When should a process stay manual?

Keep a process manual when it is rare, changes faster than the system can be maintained, lacks an authoritative source, or ends in a consequence the firm cannot test and reverse. Automation assumes someone can define correct completion.

Before building, answer five questions. What starts the run? Which record owns the input? What proves completion? Where must it stop? Who recovers a failed run? If any answer is missing, document the work before automating it.