Not a prompt list. Every automation shows what it replaces, the exact prompt, a real sample of what comes back, the tools to wire it to, and how to know when you can trust it to run without you.
Source: Anthropic Economic Index. A quarter of all AI usage is already doing back-office work. This playbook is that quarter, packaged.
Each one gets a full page: what it replaces, a workflow diagram, the prompt, a real sample output, numbered build steps naming the tools, and a trust check.
Inbound lead research briefReplaces 20–30 min of research per lead
Outbound prospect list builderReplaces 2–4 hrs per 50-row list
First-touch email drafterWrites in your actual voice
Follow-up sequencerThe deals you stop chasing at touch two
Discovery-call prep sheetQuestions a lazy competitor won't ask
Call notes → CRM + next actionsThe 15 min of admin nobody does
Proposal generatorSame-day instead of day-three
Weekly ops & pipeline reportWritten like a sharp COO
Content engineFrom work you already did
Client onboarding runnerWhere churn and refunds are born
Morning inbox triageFour buckets, drafts for the urgent
Contract & SOW generatorFrom your call notes
Support classifier & responderThe same 10 questions, handled
Candidate briefInterview questions from their resume
Social content batch schedulerA week of posts in one sitting
This is Automation 01, unedited, exactly as it appears in the PDF. All fifteen are built the same way.
A specific block of hours, quantified. If it doesn't replace real work, it isn't in here.
Trigger → Claude → output, so you can see the shape before you build it.
Copy-paste ready, with the variables bracketed. No "act as a world-class" filler.
What actually comes back, so you know whether yours is working before you trust it.
Zapier, HubSpot, Gmail, Stripe, DocuSign, Notion — the stack you already pay for.
How to tell when it's safe to let it run unsupervised — and when it isn't.
Here is Automation 01 in full. If this is useful on its own, the other fourteen are the same.
The moment a lead comes in — form fill, reply, DM — Claude produces a one-page brief: what the company does, size and funding, signals they have the problem you solve, the person's role, three personalization hooks, and a recommended first move.
Run it on 5 leads you already know well. If the briefs match reality, go live. If they hallucinate company details, add "only cite sources you can link to" to the prompt.
19 pages. Designed to be read once and used for months.
Cover
One page per automation
The 4-week build order
Read the whole thing. If it doesn't give you at least one automation worth building, email hello@aira.solutions and you get your $29 back. You keep the PDF.
A 19-page PDF, delivered instantly after checkout. 15 automations, one page each, plus a "read this first" page and a 4-week build order. No upsell wall, no drip sequence.
No. Every automation works copy-paste in Claude on day one. The build steps show how to connect it to Zapier, your CRM, or your inbox when you want it running on a trigger — but that's optional and clearly marked.
The prompts are model-agnostic and work in ChatGPT or Gemini. The tool-specific build notes assume Claude in a few places because that's what we run, but nothing depends on it.
These are the automations AIRA runs internally and installs for advisory clients. Every one has a real sample output and a trust check, which is the part generic prompt packs leave out. Read Automation 01 above and judge for yourself.
That's the $299 strategy session — 60 minutes with Joel Roberts mapping your actual workflows — or a done-for-you build. The $29 is credited if you move up.
Stripe redirects you to a private download page the moment payment clears, and the link is in your receipt. Any trouble: hello@aira.solutions.
The compounding doesn't come from having all 15. It comes from the first three running reliably while you build the rest.
Get the playbook now