The AI Agency Playbook

🎥 Video lesson — video not yet migrated (source Skool video id: 9bb9b4e16c994592ae2065803289c236).

This video is the overview of the entire playbook: how a real AI agency acquires clients, what it actually sells, and how an engagement runs from the first call to a long-term retainer. Everything in it comes from real delivery - 11 small and medium businesses partnered with in the last three months, 9 operational audits run personally, plus 70+ Upwork projects. The lessons that follow break down each step of this funnel in detail.

The Market Right Now: Hype vs Reality

Social media is telling business owners that entire departments can be automated, that agent dashboards run companies hands-off, and that all of it can be delivered in days. The reality inside most small and medium businesses: at best a team ChatGPT or Claude plan, used mostly as a chat. Companies where a few team members are experimenting with plugins and skills are the top 10-15%. Most have done nothing.

This gap is your opportunity. Every business owner already knows they need to do something about AI - the FOMO is doing your awareness marketing for you. What they don't know is HOW it applies to their business. Your job is to show them.

Why Businesses Are Stuck: The 4 Blockers

  1. Data privacy and security. AI needs to see business data to be useful. Larger companies and sensitive industries have governance and red tape around where data lives and who can see it.
  2. Data organization. Useful AI needs easy access to the data in their CRM, accounting, document and content tools. Modern tools with open APIs or MCP servers make this easy. Legacy or closed tools make it nearly impossible.
  3. No SOPs. Most businesses have zero documentation on how they actually do things. You cannot train AI on a process nobody has written down. This is why the audit exists.
  4. The training gap. These tools today are like Excel in the early 2000s. You can build someone a great spreadsheet, but if they can't navigate Excel, they can't use or edit it. Same with plugins - the team needs basic AI fluency, and the bigger the company, the more change management is required.

The tech to build everything you see online already exists. Humans are the bottleneck. That is why this is a years-long adoption wave, not a weekend trend - and why an agency that knows the process wins.

The Funnel

Acquisition: Meta ads + Upwork. Meta is the scalable channel (6 of the last 11 clients). Upwork brought 3, organic 1-2. No heavy awareness work needed - just show business owners concretely how AI applies to their company.

Discovery call. The key job here is segmenting by business size, because the offer changes:

  • Small businesses ($500K - $3M revenue): Audit at $3-5K. Below $500K, don't take them as clients - teams that small should learn the skills themselves. For this segment, a plugin-first approach is also being tested: skip the audit and sell one high-ROI plugin ($6-12K) so they see value fast. Quoting plugins for home services and construction are a great example - teams waste 5-40 hours a week on manual quoting, and cutting that by 80% pays for the plugin many times over.
  • Medium businesses ($3M - $100M revenue): Audit at $10K fixed price (and it can be priced higher once you have case studies). These companies value the consulting piece, expect change management, and anything over $2M revenue should start with the audit.

Billing: 50% upfront, 50% on completion.

The core pitch: everything you sell is ROI-funded. You are operating on the client's existing operational waste, so done right, the engagement pays for itself. You sell one of two outcomes: efficiency (reduce time on tasks) or revenue (more leads, better qualification, nurturing, reactivation).

The Audit: 4 Deliverables

The audit is 5-30 hours of interviews with key process owners, then four deliverables:

  1. Full SOPs. Visual process maps (BPMN diagrams) of every process - sales, onboarding, quoting, fulfillment, payroll. Who does what, with which tools, and how long it takes. As granular as possible, including the decisions people make subconsciously.
  2. Findings. Pain points and opportunities extracted from the SOPs, linked back to the interview transcripts.
  3. Hidden costs. Quantified inefficiency. Example: one staff member spending 40 hours a week copy-pasting between Excel sheets, at an average $50/hour rate. This shows where the biggest problems are - and where to start.
  4. AI Blueprint. The foundation plan for becoming an AI-first business: 5-30 recommended initiatives across the five offer types below, plus a technical appendix on implementation.

The 5 Things You Sell Post-Audit

  1. Plugins (60-80% of the work). AI buddies for a specific process, built in Claude Cowork, human in the loop. Fast to build (delivered in 1-2 weeks), no data storage headaches since team members authenticate their own tools, editable by the client. Priced $4-20K with returns that can hit six figures a year for larger teams.
  2. Automations. Human OUT of the loop - usually moving data from A to B with no input needed. Hosted in n8n, Zapier, or whatever the client already uses.
  3. Data migration. Flagged when a business has no single source of truth or sits on a closed legacy tool with no API or MCP. Usually a longer-term blueprint item, not an immediate sell.
  4. Custom builds. When custom software is cheaper than SaaS or no SaaS tool fits. Example: a manufacturing team with gloved hands fighting unnecessary ERP fields in Microsoft Business Central.
  5. Training and workshops. Getting a 40-100 person team fluent in AI tools is the Excel-fluency advantage of the 2020s. The compounded efficiency gain across the org is the ROI.

The Engagement Arc: Audit, First Plugin, Partner

Step 1 - Audit. Everything above.

Step 2 - First plugin. Deliver the one plugin with the most obvious ROI. Building it is the easy part - adoption is the hard part. The team isn't in Claude yet, traditional businesses have run on Excel for 20 years, and staff often fear AI is taking their jobs (it isn't - a human still runs every plugin). Budget 5-30 hours for adoption depending on the team. Once they use it and feel the ROI, they get the AI bug, they can finally separate the social media noise from real implementation, and working with them becomes dramatically easier.

Step 3 - AI partner. With trust and proven ROI, you become their long-term partner. As AI evolves, you build every new initiative with them on a retainer. This model is still being refined (Cowork and plugins are a six-month-old concept), but if you delivered steps 1 and 2 properly, this relationship does not go anywhere.

What's Next

The following lessons walk through each step of this funnel in order: traffic, sales calls, the proposal, the audit, and the post-audit retainer and plugin offers.

Next →