A handbook for everyone on the team.
Your company already has a way of doing things. Put it in a handbook that people and AI employees can both learn from.
Start with what your team knowsWhat we know. How we work.
How we do the work
Standard operating procedures · reviews · handoffs
Our customers
Product knowledge · support standards
Working here
Who to ask · where things live · the policies behind the work
Start with the questions people already ask.
A new colleague needs context before they can be useful. So does an AI employee. Your policies, product knowledge, and team conventions are a natural place to start.
“What’s the vacation policy?”
If you connect an AI employee to Slack, give it the handbook context to help with questions like this. The answer should come from your company’s policy, with a person to ask when the policy leaves something unclear.
Begin with pages you already have. Add the answers your team keeps explaining: how to submit expenses, where product information lives, or who owns a decision.
Write the work down the way you’d teach a colleague.
A standard operating procedure is a description of how your team does a job. What starts it? What should someone check? What does a good result look like?
Those instructions belong beside the company knowledge they depend on. People and AI employees can work from the same procedure, with responsibilities that suit each of them.
Real work does not run straight through. Write the checks in too: when step three fails, the procedure says go back to step one, and when step six fails, it says go back to step four. That is what turns a list of steps into the same result every time.
An SOP pays off ten seconds after you write it. You do not wait for a new hire or a resignation to see the value, because an AI employee can pick the procedure up and start running it immediately.
Preparing an expense for approval
Example procedure · Finance
- 1.
Collect the submission
Pull the receipt, amount, date, and stated reason off the request.
- 2.
Match it to a policy line
Find the category in the handbook and read the limit that applies to this team.
- 3.
Check the receipt is complete
Vendor, date, itemized total, and payment method are all present.
Something is missing. Go back to step 1, then run forward from there again.
- 4.
Compare the amount with the limit
Under the limit, note it. Over the limit, name the exception the policy allows.
- 5.
Write the summary
Amount, category, the policy line it sits under, and any exception claimed.
- 6.
Send it to the approver
The person who owns the budget makes the call.
They ask something the summary does not answer. Go back to step 4, then run forward from there again.
Decide where the work changes hands.
Once the instructions are clear, choose who does the work and where it needs a review. An AI employee might prepare an expense summary while a person keeps the approval decision.
Prepare
The AI employee checks the request against your handbook.
Review
The assigned person receives the summary and decides.
Continue
Approved work moves on; questions go back for clarification.
Handbook pages explain the work. Configured steps connect the actions, reviews, and waits; Wallfacer calls that process a playbook. When work must wait for approval, we help configure that step during setup.
Let the handbook grow with the work.
Start small. When a question comes up twice, add the answer. When a review catches a missing check, update the procedure.
Every page keeps its revision history. AI employees can read the relevant instructions and search for more context as they work.
A change passes its tests, but no one checked the screen on a phone. Add the missing check to the handbook.
- Before requesting review:
- Added line: Run the changed flow in the Simulator.
- Added line: Attach a phone-sized screenshot to the pull request.
- Added line: Describe anything you could not verify.
We’ll build the first pages with you.
Bring the handbook you have, or just a job you want help with. We can work alongside your team to write the instructions, connect the right tools, and walk through the process together.
Use a starting process, adapt it, or bring your own.
Plan your first AI employee