Agentic OS Onboarding Playbook

How to use this document: this is yours to read at your own pace — there's no scheduled session where we walk you through it. Work through it in one sitting or over a few days, in whatever order is useful. Section 0 tells you when to start; the rest builds from there.

Version 1.0 · Effective 2026-07-09 · AI Automation. This playbook is a companion to the Master Terms & Client Acknowledgment, which governs the commercial and legal relationship. Where the two differ, the Master Terms control.

0. Sequencing — your core sales asset comes first

Everything below is secondary until your website — or whatever is the primary place customers land — is live. The Agentic OS dashboard, your agent roster, and everything that runs on top of it needs something to run against. Your site is the supply-side foundation; the platform is the engine that operates on top of that foundation once it exists.

Practical rule: if your site isn't live or in final review yet, read Sections 1–3 now for context, but don't expect Sections 4–5 to make full sense yet — come back to them once it's live.

1. Before you dive in — questions worth answering for yourself

You don't need to send us answers to these. They're here so you walk into the platform with a clear picture of where you're starting from — there's no wrong answer to any of them.

A. Your AI experience today

1. Which AI chat tools have you actually used — and for how long?
2. What have you used them for: brainstorming, writing, planning, outlining a business, generating images, writing code, something else?
3. Has an AI tool ever done something for you beyond writing text back — booked something, sent an email, posted to social, touched a file on your computer? If so, how did it go?
4. On a scale of "I mostly copy-paste answers into a doc" to "I've chained multiple AI steps into a workflow" — where are you?

B. The agent platform underneath your dashboard

5. Have you taken any course or structured training on the underlying agent platform your dashboard runs on?
6. Do you understand, in your own words, what a "workspace" or "venture" is and why your business might have more than one?
7. Have you ever created, edited, or enabled/disabled an agent yourself?
8. Do you know the difference between an agent's skill (what it's allowed to do) and a task (a specific thing you asked it to go do)?

C. The agent runtime that actually executes work

9. Have you taken any course or training specifically on how the agent runtime that powers your dashboard works?
10. Do you know how you'd hand it a task today, and how you'd check whether it finished?
11. Do you know what an "approval gate" is and why some agent actions wait for your go-ahead instead of happening automatically?
12. Are you aware agents can learn new skills over time, and that you — not the AI — approve a learned skill before it goes live?

D. Email / calendar / drive / social integrations

13. Do you understand what "draft-only" access means — i.e. an agent can prepare something but can't send or post it unless you've explicitly turned that on?
14. Have you connected your business accounts (email, calendar, drive, social platforms) to the platform yet? Which ones, which are still pending?
15. Do you know what OAuth is, in plain English — "I'm granting this specific app permission to act on my account, and I can revoke that any time"?
16. If an agent drafts something for your social account, do you know where you'd go to review it before it goes live, and how you'd stop it?

E. Your business, in your own words

17. In one sentence, what does your business actually sell?
18. Who is the customer — be specific, not "everyone who might want this"?
19. What's working today to bring in customers, even in a small or informal way?
20. What's the single biggest bottleneck right now — getting found, getting trusted, converting interest into a sale, or something else?

F. Website / core sales asset status

21. Is it live, in progress, or not started?
22. If in progress, what's the blocker — content, design, a decision you haven't made yet, something on our end?
23. Once live, who's the first audience you're sending to it?

G. Technical comfort

24. Have you ever used a command line / terminal for anything, even once?
25. Have you heard of or used Claude Code before?
26. If something on the dashboard breaks or looks wrong, is your instinct to try to fix it yourself, or to flag it and wait?

H. Time and learning style

27. Realistically, how many hours a week can you put toward learning this platform (separate from running the business day to day)?
28. Do you learn best from short videos, written guides, or working through it hands-on?
29. Is there anyone else on your team who should also work through this playbook?

2. What we do — and explicitly don't — teach

Read this section carefully before assuming what's covered:

Our job is to build and operate the Agentic OS platform for you. We make sure it's live, your integrations stay connected, and the dashboard components we built are working. That's the deliverable, and we stand behind it.

We do not teach, coach, or train on the underlying products we didn't build — the agent platform, the agent runtime, the local model tooling underneath it, AI agents in general, how to write a good agent task, social media strategy, go-to-market strategy, or "best practices" in any domain outside the platform's operation. Those are real skills, worth learning, but they're not something we're positioned to coach you through — we're your platform builder and operator, not a business or AI-literacy coach.

We are also not responsible for changes, updates, upgrades, or new model releases made by AI LLM providers, AI third parties, or other external vendors. The platform depends on models, APIs, and services we don't own or control. When a provider changes how a model behaves, deprecates a feature, updates pricing, or ships a new model version, that's outside our control and outside our responsibility — we'll do our best to adapt the platform around it, but we can't guarantee any specific third-party AI product will behave identically over time.

What to do instead, when you're not sure who owns a question:

• If it's "how does this button or feature work" → that's ours, ask us.
• If it's "how do I know what task to give an agent so it does what I actually want" → see Section 3 for where to learn that, and remember: the chat inside your own dashboard is a live AI conversation — you can literally ask it to explain itself.
• If it's "what should my marketing strategy be" or "how do I run this business" → that's explicitly out of scope for us — see Section 3.

2a. Scope, fees & changes to the platform

Your monthly licensing and support keeps the platform you were delivered running: it stays live, your integrations stay connected, and the components we built keep working. It does not include new work.

Any changes, removals, upgrades, or updates to the Agentic OS platform are not part of monthly licensing and support. New features, new agents or integrations beyond what was delivered, redesigns, model migrations you request, and similar work are billed separately, per project, on a 50% upfront / 50% on delivery basis. Work on a change request begins only after the change is scoped in writing and the 50% deposit is received.

The full terms — including support scope, what's excluded, intellectual-property and data ownership, acceptable use, and your responsibility for your own third-party accounts — live in the Master Terms & Client Acknowledgment, which you sign separately. This section is a plain-language summary; the Master Terms control.

3. Where to actually learn this stuff

Not exhaustive — a starting list. Treat it as a menu, not a syllabus.

AI / agent literacy (general, not platform-specific)

• The model provider's own documentation and usage guides — usually free, and directly relevant since most agent platforms sit on top of a small number of large model providers.
• Short applied courses on prompt engineering and agent-building — a few hours each, practical.
• Creators who build in public with AI agents and coding assistants — faster-moving than formal courses, good for "how do people actually use this day to day."

Running the business itself

• A lean business-plan or business-model-canvas exercise — forces the "what do I sell, to whom, why me" answers no platform can invent for you.
• Basic small-business financial literacy (pricing, margin, cash flow) — a prerequisite for judging whether any of this is actually working.

Marketing (this one matters most — see Section 5)

• Fundamentals of content marketing, basic paid acquisition, and email marketing — enough to have an informed opinion when an agent drafts a campaign, not enough to become a full-time marketer.
• Platform-specific basics for wherever your customers actually are — go deep on the two or three channels that matter, not shallow on all of them.

If you want to build or customize yourself

• Claude Code's own onboarding materials are a genuinely good, free, self-paced starting point.
• No formal course is needed to use the dashboard day to day. A course is only useful if you want to go further and build custom agents or scripts yourself.

4. The transition — from "chatting with AI" to "running a business on it"

Each stage assumes the one before it is genuinely done, not just started.

Stage 0 — Core sales asset live. Non-negotiable prerequisite (see Section 0).

Stage 1 — Orientation to the dashboard. Not a course — get familiar with what's already running: which agents are active, what "approval gate" and "needs your attention" mean, where to look when something needs a decision. Goal: you can log in and understand what you're looking at without guessing.

Stage 2 — The mental shift: chat vs. task. The core thing anyone coming from chat-only AI tools has to relearn. A chat is a conversation that ends when the window closes. A task is a delegation — you state an outcome, an agent works toward it (possibly over minutes or hours), and something comes back for your review or runs automatically within limits you've approved. This is the single biggest mental-model gap coming from ChatGPT-style use — sit with it before moving on.

Stage 3 — Feeding the platform your business. The agents are only as good as what they know about your business. This is your work, not ours or the AI's: what you actually sell, who the customer is, what tone and voice to use, what's true about your operation. No amount of platform sophistication substitutes for telling it the truth about your business.

Stage 4 — Building further, if you want to (optional). If and when you want to go beyond the pre-built agent roster — customize behavior, script something new — that tooling is available to you. This is opt-in, not part of baseline onboarding, and the skill-building for it lives in Section 3. New build work we perform on your behalf is billed per project (see Section 2a).

Stage 5 — The growth loop. Agents handle execution — drafting, triage, research, content, scheduling. You handle strategy and reading the market. Marketing is the connective tissue between the two (see Section 5) — it's what turns "the platform can produce things" into "customers show up."

5. The honest framing

This platform puts you dramatically further ahead than starting from zero with the raw underlying tools. You didn't have to figure out agent architecture, credential handling, integration tooling, or how to wire your accounts into an agent safely — that's built, tested, and running for you. Most people never get past that setup phase on their own.

But no platform — this one included — builds you a business. It doesn't create market demand, and it doesn't replace having an actual business plan that builds real supply against that demand. The tools can execute faster than you could alone; they can't decide what's worth executing. That's still on you.

And once your product or service is actually live: marketing is one of the most crucial parts of what comes next. A great product with no one hearing about it doesn't sell itself, agents or not. Don't treat marketing as an afterthought once the platform's running — it's one of the first things to get real attention once you're live.

Glossary — quick reference

• Agent — a configured AI worker with a defined role, a set of allowed tools, and a scope of authority you control.
• Task — a specific, bounded delegation you give an agent, with a start and an end.
• Skill — what an agent is permitted to do; separate from any individual task.
• Approval gate — a checkpoint where an agent's action pauses for your explicit go-ahead before it happens.
• Workspace / venture — an isolated environment for one business or brand, with its own agents, credentials, and memory.
• Draft-only access — an integration mode where an agent can prepare something (an email, a post) but cannot send or publish it without you.
• OAuth — the standard way you grant a specific app permission to act on an account you own, revocable at any time.
• Change request / change order — any new work beyond what was delivered; scoped in writing and billed per project (see Section 2a).

Appendix — reading this on your own

• Section 1 isn't a form to fill out and submit — it's a warm-up. Let your honest answers shape how much time you spend in Sections 2–4.
• If your answers in Section A show you've never had an AI take a real-world action before, slow down on Stage 2 — that's the whole ballgame.
• If Section F shows your core sales asset isn't close to live, that's genuinely fine — bookmark Sections 4–5 and come back once it is.
• Keep your own notes from Section E somewhere — they're the raw material for Stage 3, not just background reading.

Client acknowledgment

By signing below, you confirm that you have received and read this Agentic OS Onboarding Playbook, that you understand it is informational and reflects best-effort opinions rather than professional advice or any guarantee of results, and that the binding terms of the relationship are set out in the separately signed Master Terms & Client Acknowledgment.

(In the PDF edition of this playbook, the signature block below is a fillable, digitally signable field.)

Disclaimer

This playbook, and any guidance, opinions, templates, or recommendations in it, are provided by AI Automation ("we," "us," "our") on an "AS IS" and "AS AVAILABLE" basis, for general informational purposes only, without warranties of any kind, express or implied, including any implied warranty of merchantability, fitness for a particular purpose, or non-infringement.

No professional advice. Nothing in this playbook is legal, financial, tax, accounting, marketing, or other professional advice. It reflects our best-effort opinions and operational experience building and operating the Agentic OS platform — it is not a substitute for a qualified attorney, accountant, financial advisor, marketing professional, or other licensed advisor appropriate to your situation. You are solely responsible for evaluating whether any idea, template, or recommendation here is suitable for your specific business, market, and legal jurisdiction before acting on it.

No guarantee of results. We make no representation, warranty, or guarantee — express or implied — about the outcome of following any part of this playbook, including business growth, revenue, customer acquisition, marketing performance, or the success or failure of your business generally. Running a successful business depends on factors entirely outside our control, including market demand, competition, execution, capital, timing, and conditions specific to your industry and jurisdiction. Nothing here is a promise, prediction, or assurance of any particular result.

Not the only way. This playbook reflects one approach, informed by our experience, at a specific point in time. We do not represent or warrant that it is the only, best, or most current way to start or run a business, or to use artificial intelligence, AI agents, or agentic platforms generally. This field changes quickly and parts of this document may be outdated by the time you read it.

Third-party AI, models, and platforms. The Agentic OS platform, and this playbook, depend in part on third-party AI models, providers, and platforms we do not own, control, or operate. Those third parties may change, update, deprecate, restrict, or discontinue their models, features, pricing, terms, or availability at any time, without notice to us or to you. We are not responsible or liable for any change, error, outage, output, behavior, or consequence arising from any third-party AI model, update, or upgrade — including changes made unilaterally by a model or platform provider. AI-generated output can be incomplete, inaccurate, outdated, or unsuitable for your purpose, and should be independently reviewed before you rely on it for any consequential decision.

No fiduciary or advisory relationship. Providing this playbook does not create a fiduciary, advisory, consulting, or agency relationship beyond what is separately and expressly agreed in a signed services agreement between you and us. Any such relationship, and its scope, is governed exclusively by that separate written agreement, not by this document.

Limitation of liability. To the fullest extent permitted by applicable law, we will not be liable for any indirect, incidental, special, consequential, exemplary, or punitive damages, or for any loss of profits, revenue, data, goodwill, or business opportunity, arising out of or related to this playbook or reliance on it, even if we were advised of the possibility of such damages. Our total aggregate liability arising from or related to this playbook, under any theory of liability, will not exceed the amount (if any) you paid us specifically for this document.

Your responsibility. You are responsible for reviewing, testing, and approving any AI agent action, communication, or output before it is relied upon, published, sent, or acted on — including anything involving your customers, finances, contracts, or public communications. You remain the decision-maker for your business at all times.

Governing law. This playbook and any dispute arising out of or relating to it are governed by the laws of the State of Delaware, without regard to its conflict-of-laws rules. If any part of this disclaimer is found unenforceable in a given jurisdiction, the remaining parts remain in full effect. This disclaimer may be updated from time to time; the version delivered with your copy of this playbook governs your use of it.

Document control — Version 1.0 · Effective 2026-07-09 · © AI Automation. Companion document: Master Terms & Client Acknowledgment (v1.0).