SOPs: AI agent SOPs that run as steps, not wiki pages.

An AI agent SOP, or agent operating procedure, is a standard operating procedure written as workflow steps the agent executes one at a time, with stops where a person decides. ConvOps turns the SOP your team keeps in a wiki into a process the engine holds, so every run follows it and edits reach the next run.

claude · codex · cursor · any mcp client · one step at a time · a person decides at the gate

Claude Codeconnected

new chat

ConvOps · task · exampleVendor onboarding SOP
task

Vendor onboarding · example

step 1 of 5
collect check why approve approvaldone
the problem

The SOP exists. Nobody runs it.

The procedure lives in a wiki. People skim it, agents never see it, and the steps drift.

“Did anyone follow it?”

The wiki says a manager approves. Nothing shows whether one did.

“Which version is current?”

Three copies of the procedure, edited by three people, in three places.

“Does the agent know?”

Paste the SOP into a prompt and the model decides which parts matter today.

watch it work

One SOP, from wiki to running.

An example procedure, turned into a workflow from Claude Code, then run on a real case.

01you, in Claude Code

Paste the procedure.

You hand the agent the wiki page. It proposes a workflow: one step per action, a gate where a person decides, a step that must record its reasoning.

example
Claude Codeconnected to ConvOps
Turn our vendor onboarding SOP into a workflow. A manager approves before the vendor is created.
workflow created · 5 steps · 1 approval gate
Collect, check, record why, manager approves, create. Want to review the steps?
02the engine

The wiki becomes steps.

The same procedure, now held by the engine. The agent sees one step at a time, so it cannot jump ahead to the part it likes.

example
Vendor onboarding (wiki) · example
SOP · vendor onboardingworkflow
  1. Collect details
  2. Run the checks
  3. Record the reasoning
  4. Manager approvesa person decides
  5. Create the vendor
03the AI · one step

The checks land on the task.

The agent runs the checks the SOP names and writes the findings down. The step closes only with a reason recorded.

example
Vendor onboarding · examplestep 3 of 5

Checks

recorded on the task

tax idformat valid
bank detailsmatch the invoice
policycontract over threshold: needs approval
reason recordedrequire_context
04the manager

A person decides.

The run stops at the approval gate. The manager reads the findings and decides. Until approval is passed, the vendor is not created.

example
waiting for you

Approve this vendor?

Checks passed · contract over threshold · reasoning recorded

Hold itApprove
approvedStep 5 unlockedCreating the vendor was locked until now.
05the record

The record writes itself.

Every step lands in the history with its reason. Field changes on the task keep the agent label and the signed-in person apart.

example
ConvOps · task activityrecord
  • collectdetails captured
  • checkfindings recorded
  • approveapproval passed
  • donevendor created

agent label · implementer | person · verified from the credential

decisionContracts over threshold always need a manager.linked to this task
example
Claude Codeconnected to ConvOps
Turn our vendor onboarding SOP into a workflow. A manager approves before the vendor is created.
workflow created · 5 steps · 1 approval gate
Collect, check, record why, manager approves, create. Want to review the steps?
the workflow

The SOP, running.

Each step is served when the run reaches it. The approval is where the SOP says a person decides.

workflow · Vendor onboarding SOPrunning · step 1 of 5
  1. 01

    Collect details

    Legal name, tax id, bank details and contact, as the SOP lists them.

  2. 02

    Run the checks

    Against the policy the SOP names. Findings recorded on the task.

  3. 03

    Record the reasoning

    The step cannot close without the reason written down.

    decision
  4. 04

    Manager approves

    An approval gate holds the run until a person passes it.

    needs a person
  5. 05

    Create the vendor

    Only after approval. The step history shows every stage.

what it remembers

Why the procedure is the procedure.

The reasons behind each step, kept by the team and recalled when the next case arrives.

Decisions

Why the threshold is where it is, recorded at the step that applied it.

why we do it

Corrections

The check that caught a bad bank detail last month, so the next run looks for it.

what went wrong

Where things live

The policy, the vendor sheet and the contract folder. Found once, kept.

where it lives

Rules in context

Policies attached to a workflow reach the agent with the step they apply to.

guidance, not a gate

the difference

Wiki SOP vs running SOP.

  • The procedure sits in a wiki page
  • The agent reads it, or does not
  • Sign-offs happen in chat, if at all
  • Each copy of the SOP drifts on its own

Steps in order

The engine serves the current step only. Steps inside one workflow never run at the same time.

Approval is a gate

requires_approval refuses the advance until approval is passed, and the response lists what is missing.

The why is a gate

require_context keeps a step open until the reason is recorded, so the decision travels with the work.

What the record covers

Step history with reasons, an activity feed, and field-level before and after on tasks and workflow instances.

start this week

Four moves to your first run.

Connect the AI tool your team already uses, install the workflows, and run one real piece of work through them.

  1. move 1

    Pick one SOP your team repeats weekly and paste the wiki page into your AI client

  2. move 2

    Ask the agent to draft it as a workflow: one step per action, a gate where a person decides

  3. move 3

    Run one real case through it and read the step history afterwards

  4. move 4

    Fix the step that was unclear once; every later run follows the new version

install these

questions

What teams ask.

What is an AI agent SOP?

A standard operating procedure written so an AI agent can execute it: ordered steps with instructions for each, and explicit stops where a person decides. Research such as SOP-Agent (arXiv:2501.09316) shows procedure-shaped guidance helps general-purpose agents; ConvOps keeps that procedure in an engine instead of a prompt.

Why not paste the SOP into the prompt or a skill?

The model then decides which parts to follow. In a workflow, the engine serves one step at a time and checks gates before the run moves, so an approval gate pauses the run until a person approves, whatever the model judged.

Who writes the workflow?

Usually the agent. Paste the SOP and ask for a workflow; it creates the steps and gates over MCP, and you review them. There is no canvas editor: changes are made in conversation and stored by the engine.

What happens when the SOP changes?

Edit the step once. Runs that start afterwards follow the new version, for every person and every AI client on the workspace.

What happens if the approver says no?

If the approval step names an on_rejected target, a rejection moves the run to that step, for example back to the checks, and the step history records it. If the step has no target, the rejection is refused and the run stays at the approval.

Can steps of an SOP run in parallel?

Not inside one workflow: steps run in order. Independent pieces of work are split into child tasks that start together, and the parent waits for all of them.