AI Employee Operating Manual
The strategy layer behind every successful "AI employee": a deployment-ready operating manual personalised to your business — master system prompt, task playbooks, guardrails, escalation matrix, KPI scorecard and a 30-day rollout plan. Run it on ChatGPT, Claude, Lindy, Relevance or any agent platform.
You have picked the platform. What you do not have is the thing you would hand a new starter on day one: a written account of what the role owns, where it must stop, and what a good week looks like. The AI Employee Operating Manual is A$129 one-off, and it fills that gap for one nominated role — an AI SDR, a support agent or an EA — built around your business, what you sell, your market and the goal you want the role to hit. Turnaround is around 15 minutes, and it lands on your order page: readable in the browser, downloadable as Markdown, printable to PDF if you want it in the team folder.
Authority first, then the prompt
The role charter comes first, because it is the cheapest document to write during setup and the most expensive question to answer afterwards: what the AI owns outright, and what stays with a person.
The master system prompt follows, written around your business, your market and the role you nominated, ready to drop into your platform's instruction field. Beneath it sit five to seven task playbooks, each with worked examples and a quality bar, so an acceptable output is distinguishable from a poor one. What arrives is a document rather than a configured agent — the pasting, connecting and running stay with you.
Guardrails, and the point where a person takes the call
Pricing, claims, customer information, anything that needs a person's name attached to it — the guardrails and data rules name the things an AI should never improvise. Each is paired in the escalation matrix with a defined handover: what triggers it, who catches it, and what the AI says while it waits.
The KPI scorecard sets weekly targets so reviewing the role becomes a short check rather than a debate. It is a written framework, not a dashboard. It does not connect to your CRM or inbox and measures nothing on its own; the numbers come from wherever you already track them.
Written so a change of platform costs you nothing
Nothing in the manual is tied to one vendor. The prompt and playbooks are written to run on ChatGPT, Claude, Lindy, Relevance or whichever agent platform you settle on, so swapping tools later does not send you back to first principles. The 30-day rollout plan sequences the opening month — what to switch on first, what to keep supervised — and the prompt-maintenance guide covers editing the instructions as the role beds in. Four free AI revisions come with the order, which is usually enough to absorb a change of direction after the first fortnight of real output.
What you get
- Role charter — what the AI owns vs what stays human
- Master system prompt (paste-ready, personalised)
- 5–7 task playbooks with examples and quality bars
- Guardrails, data rules and escalation matrix
- KPI scorecard with weekly targets
- 30-day rollout plan + prompt-maintenance guide
- 4 free AI revisions included
Turnaround: ~15 minutes. Delivery: On-page + Markdown download (print to PDF).
Frequently asked questions
My platform splits instructions across several fields rather than one box. Does that break it?
No, but it does mean a little more work at the paste stage. The prompt and playbooks are plain text and platform-agnostic, so where a tool wants the role description, the rules and the examples in separate fields, you move the relevant sections into the right boxes instead of dropping in one block.
I want an SDR and a support agent. Is that one manual or two?
Two. You nominate a single AI role, and the charter, system prompt, playbooks and KPI targets are all built around it. A second role is a separate order with its own inputs, so it gets guardrails and a scorecard suited to that job rather than a compromise between the two.
Do you configure or run the agent on our side?
No. This is a written manual, delivered as a document you read on your order page and download as Markdown. Nothing is installed, connected or monitored by us. You paste the prompt into your own AI platform and run the role yourself.
The role will change once it is live. Can the manual keep up?
That is what the four free AI revisions and the prompt-maintenance guide are for. Once you have watched the role work for a week or two, use a revision to tighten an escalation trigger, rewrite a playbook for a different customer type, or reset a KPI target that turned out to be wrong.
Will the scorecard tell me whether the AI is actually performing?
It gives you the targets and the structure to judge against, which is not the same as measuring. The scorecard does not track, report or connect to your CRM, inbox or phone system. You bring the numbers from the systems you already use and score the role against the weekly targets it sets.