How to Build & Sell AI Automations: Ultimate Beginner's Guide

Introduction

In a world being transformed by AI, one skill stands above all others: AI automation. Master this and you won't just survive the AI revolution—you'll thrive in it. I'm living proof. Two years ago, I taught myself to build no-code AI automations without prior experience. Since then, I've built multiple AI businesses, generated millions in revenue, grown this channel to over 500,000 subscribers, and built AI systems for major global brands. Learning how to build AI automations completely changed my life. In this full course, I'll teach you everything I've learned about building AI automations and making money with them—even if you don't know how to code. AI won't replace you; the person using AI will. My hope is that this video helps you learn a powerful skill to build the life you want before it's too late. As you can tell from the length of this video, I'm not holding anything back.

I've split it into three chapters: first, foundations—what AI automation is, types, how it works, and core concepts—no technical background required. Second, we'll build real automations together, including cutting-edge voice agents. Third, I'll share my blueprint for monetising AI automation skills as this technology explodes, including the strategies I used to generate millions.

If you're new here, why I'm qualified to teach this: my name is Liam Ottley. Two years ago I started learning AI with no prior experience, teaching myself to build AI automations and chatbots and documenting the journey on this channel from day one. That led to Morningside AI, my AI automation agency, where we've built systems from basic customer support to full AI SaaS platforms for major brands. I also run an AI SaaS called Agentive with over 70,000 users. We've worked with publicly traded companies and even an NBA team, and I run the world's largest AI automation and business community with 180,000+ members on Skool. Through that community and this channel, I've taught hundreds of thousands of people how to build and monetise AI automation. Everything I'm about to teach helped me achieve this. Let's dive in. There's a lot to cover, and I don't want you to give up halfway—so let's get clear on why learning AI automation is one of the most valuable skills of the next decade.

Why Learn AI Automation?

Whether you're a student, employee, or entrepreneur, here are truths about AI and jobs. McKinsey predicts AI and automation could replace up to 50% of current work activities by 2030, and the World Economic Forum says 41% of companies plan to reduce staff due to AI. That sounds bleak, but the same reports show opportunity for those willing to act: 50% of employers plan to reorient their business in response to AI, and 66% plan to hire talent with specific AI skills such as AI workflow automation. On one hand: major automation in the next 5–10 years. On the other hand, most employers are searching for people with AI skills or at least AI literacy.

Why? AI-literate people who can spot and implement automation opportunities can achieve 5–10× the output of those who can't. Brushing up on AI to get on the winning side is easier than you think—watch this video to build your skills base. If you doubt that a little self-study goes a long way, consider this clip from the All-In podcast with Naval Ravikant. His advice: brush up on AI, learn a little, tinker, then re-apply for the job that once rejected you—watch how they pull you in. This video is exactly what he's talking about. Whether you're a student aiming to stand out, an employee becoming irreplaceable, or an entrepreneur scaling with cutting-edge tools, this is for you. Close your tabs, grab a notebook and pen, and commit to finishing this training so you're empowered by AI, not replaced by it. If you've done that, let's get started.

What is AI Automation?

Step one is knowing what automation actually is. The AI part is new; automation has been around a long time. In simple terms, an automation is a system that does a task for you without you lifting a finger—like setting up a robot to handle repetitive work. Old-school automations, long before ChatGPT, were often built on platforms like Zapier and used for years by small and medium businesses. Examples: saving website form info to a spreadsheet or alerting Slack when an email arrives—if this, then that. No thinking—just doing. The benefit: frees people from basic work and saves businesses from hiring for tiny tasks.

Then ChatGPT launched in late 2022 and blew the field wide open, turning automation from a niche trick into a mainstream revolution. Generative AI added to automation was like putting a V12 on a bicycle: these models handle trickier tasks that used to require a human brain. Instead of just updating a spreadsheet row, platforms like Make.com can now use ChatGPT to write LinkedIn posts in your voice, extract names and numbers from huge documents, classify emails, summarise long text, identify items in images, and even create new images or videos from a few words.

ChatGPT and other generative tools gave us human-level intelligence on demand. Think of them as a super-smart friend who can do almost anything you ask—if your instructions (prompts) are clear. Using automation platforms, we can wire these "friends" into systems in thousands of ways: pick the right tool, give a clear prompt, watch the magic. That's how the AI automation industry was born. Since the field is new, here's a simple working definition: an AI automation is a system that uses AI to automatically do complex tasks that normally require a human. The big difference from old-school automation is the kind of tasks now possible—reasoning, creativity, and decision-making. You're learning to build digital workers that do powerful things without you lifting a finger. That's why people are racing to learn this—to stay ahead. You can tailor automations to any job or workflow: students organise notes; employees offload paperwork to shine on bigger projects; entrepreneurs run business tasks while they sleep—at far lower cost than hiring, no breaks or holidays required. This is why anyone who can build these systems becomes instantly more valuable. Imagine automating email sorting or scheduling and reclaiming time for what matters.

The AI Automation Landscape

AI automation is a broad term covering many systems due to rapid advances in AI agents and AI tools. To make sense of it, here are three categories I use:

Conversational AI: systems that chat with people—website chatbots or voice agents that handle back-and-forth conversations once done by humans.

AI tools: systems that do a specific job on request to help workers get more done—for example, take a blog link, scrape the page, research the topic, then use ChatGPT to draft a new version.

AI workflow automations: systems that run a series of tasks by themselves when triggered or on a schedule, using AI to make decisions once requiring a human. Example: call customers 14 days after purchase via an AI voice agent to request feedback and a review—fully automated.

We'll build three automations that integrate these types so you can see them in action. When people say "AI automations," they usually mean workflow automations—a chain of steps using AI in various ways to accomplish tasks. This last category is the most powerful because it can incorporate agents, conversational AI, and tools to create end-to-end processes more valuable than the parts alone. What you're about to learn is the foundational skill to build systems that save time, make money, and make life easier—at school, work, or in your own business. After years in the game, it's one of the most valuable skills I've ever picked up.

Anatomy of an Automation

Now that you understand what AI automations are, let's look under the hood and see how they work. I've been breaking down complex AI topics for years, and I'll make this easy to understand. Think of an AI automation like a factory assembly line: different stations work together to build something from start to finish. It's a team of robots, each with a specific job, passing the project along until it's done.

There are five key parts. Trigger: the first step—the factory's start button or whistle that says "let's go." It kicks everything into gear. It might be a new email in your inbox, a website form submission, or a specific time of day (a schedule). Filter: not everything that starts the automation should keep going. A filter checks if what came in is the right stuff to work on—like quality control on a production line. If it's not good enough, it's tossed; if it is, it moves forward. Actions: where the real work gets done. These are the steps your automation takes—sending an email, updating a list, creating a report—often several in sequence, like a product moving down the line. Intelligence layer: where the AI shines. It's a smart station that can think, analyse, and decide on the spot, guided by prompting. It looks at each task, figures out what's needed, adapts to context, decides urgency, writes a custom message, or extracts key info from messy data—going beyond preset rules. Formatter: data often needs adjustments along the way; a formatter prepares things for the next step. Output: the finished product. Your automation delivers the final result—a message to your team, an updated file, or a completed document.

At the end of the line, everything comes together—like pulling a pizza from the oven: hot, ready, done. Here's a simple example most jobs can relate to: an automation that handles incoming customer emails. Trigger: a new email lands in the support inbox. Filter: check if it's important—if it mentions "urgent" or "problem," pass it along; otherwise, end the sequence. Intelligence layer: use AI to read the email, identify the topic, and draft a helpful response. Here, the intelligence layer can also act as a formatter, packaging the response in the required format. Actions: send the reply and alert the boss on Slack if needed. Output: log everything neatly and mark it as handled.

These same building blocks scale to much bigger processes, which is why AI automations are a game-changer. They combine these parts to create smart systems that save time and boost efficiency—whether you're juggling tasks at work, at school, or running a business. Now that you have a handle on what AI automations are and how they're built, let's look at the tools that make this possible.

Exploring the Ecosystem

Don't worry if this sounds techy. These tools quickly become second nature. Creating automations starts with choosing a main automation platform—a workflow builder. It's the command centre of your automation factory, a blank canvas to design with the building blocks above. Some of those blocks are powered by AI like ChatGPT. Popular workflow builders include Make.com (used later in the tutorial), Zapier for quick setups, and n8n for greater control. They're the brain of your operation, controlling how everything fits together.

Workflow builders connect to other tools to get the job done: databases and spreadsheets (Airtable, Google Sheets) for storing and structuring information—the filing cabinets where data is saved and retrieved. Communication tools (Slack, Gmail) for passing information around automatically. AI models (OpenAI's ChatGPT) for human-level reasoning—a "genius buddy" that solves problems when given clear instructions. Scheduling tools (Calendly, Google Calendar) to handle time and meetings—the personal planner to keep things on schedule. Forms and intake tools (Typeform, Tally) to collect information—common input triggers for automations.

As your automation runs, it's like a group project where each connected app contributes its specialty to move the task forward. By building a workflow, you act as the factory boss. Once you know what each tool can do, you mix and match them to work smoothly—like snapping LEGO pieces together to make something useful. You decide which tools to connect, in what order, and how AI makes the system smarter—choosing models (Gemini, OpenAI) and search tools where needed. It's creative problem-solving: you know point A and point B; AI helps you bridge them.

Examples: a student builds a study organiser that summarises lecture recordings, creates flashcards, and schedules reviews around exam dates. An employee builds a meeting assistant that records, transcribes, generates action summaries, and updates project tools. An entrepreneur designs a lead-qualification system that qualifies leads, calls them with an AI voice agent, and sends a custom proposal. That's exactly what you'll build next: a lead-qualification automation, step by step, starting simple and getting more complex. We focus on this business workflow because it offers strong opportunities to monetise your new skills—covered in depth at the end. Stick with me to learn how to start earning immediately.

Foundations Recap

Quick summary of this section: an AI automation is a system that uses AI to automatically perform complex tasks that used to require humans. We build them in workflow builders that integrate with many other tools. Every automation has six components: trigger (what starts the workflow), filter (conditions that must be met), intelligence layer (AI that processes info and makes decisions), actions (the tasks performed), formatters (clean things up), and output (the final deliverable). If anything is unclear, rewatch the earlier sections and be ready for the build phase. It's important to understand this foundation before we start the technical work next. I've layered the information intentionally; make sure you understand everything here before moving on. If that's all good, let's look at what we're going to build together.

With the foundation in place, we're moving into the second chapter to build three AI automations from scratch. We'll start with a beginner-friendly build and work up to something more advanced and valuable. Each automation builds on the previous one—you must complete the first to make the second, and so on. Trust the process; the progression is intentional. Over the next chapter, you'll learn the key skills to start building your own AI automations and tap into the opportunity in AI automation.

The system we'll build over three sections is an AI-led qualification and proposal generation system. In tutorial one, we'll auto-qualify leads after they submit a website form. In tutorial two, we'll add an AI voice agent to call the lead for more information. In tutorial three, we'll auto-generate proposals for qualified leads using info from the call and the form. The point is immediate momentum: as soon as someone is interested, they receive a proposal quickly.

Qualification is crucial at scale because not everyone is a fit. For example, an accountancy firm might only work with doctors—if a builder applies, that's not a qualified lead. Many businesses also need to create proposals for qualified leads, but most proposals do not convert—often ~20% at best—so manual effort here is a major time sink. What we're building solves multiple problems common to most businesses and is valuable enough to sell directly once you're done.

Without AI automation, a human rep must: check new submissions, review details, evaluate fit, research the company, make calls, deliver pitches, and manually create custom proposals—hours of repetitive work each week and a risk of leads slipping through the cracks due to mis-qualification, poor research, or slow responses. Faster response times significantly boost conversions.

With AI automation, we transform this into an efficient, scalable workflow: lead submits form → system qualifies with AI → system researches the company with AI → system places an outbound AI voice call to pitch the offer → system saves the call outcome, summarises the conversation, and generates a personalised proposal—all without human intervention. This is the power of AI automation: we turn hours of manual work into autonomous workflows, built with popular tools from the automation ecosystem. All resources, templates, and prompts are available free in the Skool community linked in the description; request to join, get accepted, and search for the video title to find everything. Let's get into it.

What We're Building

Before we jump in, it's important to understand what we're doing in builds one, two, and three and how they fit together. Being thrown terms like Airtable, Slack, and Make can be confusing, so here's a clear orientation of what we're building and why. We're creating an AI qualification system for inbound business leads. Think of a website form—Book a Free Consultation—via ads or organic traffic. Some submissions won't be a fit for your services; qualification prevents wasted calls.

You'll build a powerful system that starts basic, then becomes more advanced and sellable. In Build 1, we'll create a Tally form and store responses in Airtable. Make.com will watch Airtable for new records and trigger the automation. We'll leverage Airtable's built-in AI for immediate qualification instead of calling an external LLM. A custom AI field will analyse company, budget, and needs, then output Qualified/Not Qualified based on your prompt. If Qualified, Make sends the lead an email to book a call and posts to Slack for the sales team. Using Airtable AI here simplifies Make: the watcher only passes rows where the qualification equals Qualified. This is Build 1: immediate AI qualification in Airtable, automatic email to the prospect, and Slack alert to sales. Before each build, I'll add quick context so you're 100% clear on what we're doing.

Build #1

We need to gather lead information. We'll use Tally, an easy-to-use form builder. Click New Form, name it How can we help?, and build it from scratch using blocks: Short Answer for first name and last name (with labels), Email, Phone Number, Short Answer for company, Number for budget, and Long Answer for needs. Customise appearance (background, text, button, accent), Publish, copy the share link, and preview the form. Use as a landing page or embed on a site.

Store responses in Airtable—think Google Sheets on steroids; it stores and processes information and can even power lightweight apps. Create a Base called Lead Base. Add fields matching the form: First Name, Last Name, Email, Phone, Company, Budget, Notes/Details, Created On, and Qualification. For simple qualification, we'll mark leads as qualified if budget ≥ $10,000 using Airtable's AI Assistant to create an AI field Qualification that sets Qualified/Not Qualified based on the Budget field. If you get errors saving the AI prompt, verify field names and references. Create another AI field to generate a descriptive sales message. Add a Date field Contacted On (Make will write to this at the end of the workflow). Clean up: rename the sheet Lead Contacts, and the record type Lead.

Connect Tally → Airtable via Tally's integration: name the connection, select the base/table, and map fields. Test with dummy data (budget ≥ $10,000). You should see the new record in Airtable, auto-Qualified with the AI message populated. Now build the workflow in Make.com. The dashboard shows scenarios, usage, and errors. Create a new scenario. Add Airtable – Watch Records and connect using a token from Airtable's Developer Hub (keys are deprecated): allow read/write, copy the token, and paste into Make. Select the Lead Base and Contacts table; set the Trigger field to Created On so Make only acts on new entries. In Formula, filter records where Qualification = "Qualified". Save, choose Start from now on, and Run—you should see the qualified record. Note: because we use Created On, each test requires a new lead.

We need two actions: send an email and notify Slack. Insert a Router to branch these in parallel. Gmail branch: Add Gmail – Send an Email and connect your account. For personal @gmail.com accounts, create OAuth creds in Google Cloud: new project "make"; Enable APIs → Gmail API; OAuth consent screen (External), app name "make," contact email; add Gmail scopes for read/compose/send/drafts/metadata/labels; Authorised domains: make.com, integromat.com; Test users: add your Gmail; Credentials → Web application, add Make's redirect URI, then copy Client ID/Secret back into Make. Sign in and grant permissions.

Compose the email: subject uses the lead's name; body thanks them and links scheduling. Create a Calendly 30-minute Intro Call with Google Meet, set availability, add a short description, copy the booking link, and paste into the email; set recipient to the lead's email. Slack branch: Add Slack – Create a Message. If you don't use Slack, you can replace this with another Gmail notification to your team. To show Slack setup: create a workspace (free), add a #marketing channel. Connect Slack to Make and select #marketing. Use the AI message from Airtable as the content. If it looks messy (a collection), insert a Formatter – Text parser with a regex to extract just the message value. Feed that clean text into Slack.

Update Airtable: After Gmail, add Airtable – Update a Record to write Contacted On = now() for the same record. Verify the timestamp appears in Airtable. Turn on Scheduling to run automatically (e.g., every 15 minutes). Save often; you can revert versions if needed. The scenario view shows apps/modules used and helps with troubleshooting. That's Build 1: immediate AI qualification in Airtable, automated email invite via Gmail/Calendly, and a Slack alert—clean, minimal, and effective. Up next in Build 2, we'll add a voice agent. There are two types: inbound (people call the AI) and outbound (the AI calls people). We'll build outbound to gather more info from qualified leads. Because the form collects minimal data (by design to maximise submissions), the voice call expands qualification—asking about the business, needs, and goals—and handles unanswered calls, with outcomes recorded in Airtable. Build 2 layers deeper qualification on top of Build 1, ensuring sales only talk to the best-fit leads.

Build #2

We're now adding AI voice calls to qualify leads more effectively. We already have a strong foundation: a system that identifies qualified leads, follows up via email, and notifies your team. That alone puts you ahead of the curve. But what if we could make the system even more responsive, conversational, and human? That's where an AI voice agent comes in. By weaving it into our workflow, we can automatically call leads, gauge their interest through real conversation, and tailor our follow-up based on their responses—all without lifting a finger. This is where automation becomes more than a task runner; it becomes a teammate. It's the closest thing we have to human conversation at scale, and when it comes to leads, timing and tone matter. Research shows that responding to leads immediately increases success rates by 7 to 9x.

In this section, we'll integrate Vapi, a popular AI voice agent service, into our workflow. First, we'll set up our voice agent. Then we'll configure Make to have it call our leads. Finally, we'll enhance its effectiveness by feeding it custom research about each lead, enabling more targeted pitches. So what exactly is an AI-powered voice agent? Think of it as Siri or Alexa, but designed for natural phone conversations. It can make calls or answer them and do things like follow up with potential customers, schedule appointments, answer FAQs, collect feedback, or conduct surveys. Platforms that offer these include ElevenLabs (for highly realistic voices) and Vapi (known for speed, affordability, and beginner-friendly integration). We'll use Vapi because it's cost-effective and integrates easily with other tools.

Create your Vapi account and go to the Assistants tab → Create Assistant. You can start from a template (like "Lead Qualification Specialist"), but we'll make our own from scratch and name it Ben. When choosing which provider and model to use, balance ability, speed, and cost. The response time is called latency—you'll see it listed beside each model. For now, we'll go with GPT-4.1 Mini, the fastest and cheapest at the time of this recording. The prompt defines your agent's behaviour and objectives. We'll write one that includes: Its identity and role: a voice assistant representing Edge AI, an AI automation agency. Its goal: to pitch services and determine whether the lead wants a proposal. Its tone: knowledgeable, curious, not pushy—using natural language ("uh", "mhm") for realism. Its structure: a flow that introduces services, asks about needs, acknowledges responses, offers a proposal, and politely closes.

Keep your prompts concise—too many steps can confuse the agent. If your use case is more complex (like customer support), you can upload files to its knowledge base, but note that this increases latency. The simpler the setup, the more responsive your agent. Before publishing, configure a few extras: Voice: choose from multiple styles; we'll use Elliot. Background noise: optional, but adds realism and masks small AI pauses. Tools: these allow the agent to send data or trigger workflows, but we'll leave that untouched for now. The call summary is where Vapi automatically transcribes the conversation and generates a summary. We'll customise its built-in prompt to include details that will help us generate a proposal later on—two to three sentences summarising goals, tone, and any key details.

Next, we'll instruct Vapi to judge whether the call was successful—based solely on whether the lead wants to receive a proposal. This pass/fail result (true or false) will be used in Make to decide next steps. Finally, toggle on voicemail detection (provider: Vapi) to track if a call was answered or not. Once done, click Publish. Perfect—our assistant works as intended. It's now the voice of our outreach system, ready to make calls automatically.

Let's walk through the new lead journey. The trigger remains the same: a new qualified Airtable record kicks things off. Then the voice agent calls the lead using their number. We'll insert a brief pause (300 seconds) to give the conversation time to complete before analysing results. Back in Airtable, we'll add three new fields: Summary (long text), Interested (checkbox), and Proposal Sent On (date field, used later). In Make, clone the previous "Lead Qualifier" scenario and rename it Lead Qualifier + Voice Agent. The first step still watches Airtable for new leads. Then add the Vapi module to make an outbound call. Connect it using an API key from the Vapi dashboard. If you don't have one, create a Private Key and store it safely. Paste it into Make to connect.

Configure the call with: The Assistant ID (copied from your Vapi assistant page). The Lead's phone number (from Airtable). The Outbound number ID (from Vapi's "Phone Numbers" tab; US area codes only for free). Add a sleep module (300 seconds) to allow time for the call. Then test by duplicating an Airtable record to trigger a new run. Once the call completes, the automation becomes observant—it now evaluates what happened. Did someone pick up? Did the assistant succeed? Was the lead interested? To check that, we'll use an HTTP module to fetch the call record from Vapi's API.

APIs act like structured order forms between software systems. Just like a factory requests materials using an official form, Make uses an API request to ask Vapi for call data. We'll send an HTTP GET request to Vapi's call endpoint with the Call ID and API Key (using a header like Authorization: Bearer [key]). Enable parse response to structure the data for later use. Once run, we receive call data including the summary and analysis (success = true/false). We'll use these to update Airtable records automatically.

Add a router in Make to handle answered vs. unanswered calls. Answered (where ended_reason = customer_ended_call). Unanswered (where it does not). For answered calls, add another router for interested vs. not interested leads, based on the analysis result (true = interested). For both, use Airtable → Update Record: Write the call summary, Tick the Interested box (true/false), Update the Contacted On date. For unanswered calls, reuse the email + Slack steps from Build 1 to follow up and alert the team. Run the scenario again to confirm everything works.

Recap: We now have a system that detects qualified leads, places an AI-powered voice call, waits, listens, and reacts—logging outcomes in Airtable. But the assistant can get even smarter. Right now, it doesn't know the lead's name or company context, so its pitch is limited. Next, we'll use OpenAI to research leads and pass that intelligence to Vapi for personalised calls. We'll fetch web data using OpenAI's search-enabled model, summarise findings, and include them in the call variables—so the agent can reference them during the conversation.

Since the current Vapi module can't accept custom input (as of this recording), we'll replace it with an HTTP POST request to Vapi's API. The setup mirrors before: URL, headers, and a JSON body specifying assistant ID, phone number, and an "assistant override" section for dynamic data—like the lead's name, company, and research summary. JSON (JavaScript Object Notation) is just structured data: key–value pairs. We'll format our request body in JSON and set Content-Type: application/json. In the assistant override, we'll pass custom variables. Then, back in the Vapi dashboard, edit Ben's prompt to expect and use these variables to personalise the call.

To perform the research itself, we'll add an OpenAI – Create Chat Completion module before the Vapi call. Connect your OpenAI account, choose a search-capable model (like GPT-4o-mini Search Preview), and write a prompt that asks it to summarise how an AI automation agency could help that company. Test the module—it performs the research, but outputs paragraphs. Since line breaks can break JSON formatting, we'll use a Text Parser (Replace) to remove line breaks via regex. Set it to global match. Feed the cleaned text into the company_research field in your JSON body. Finally, re-link your HTTP modules: since the old Vapi module is gone, reference the call ID from the new one. Test the new workflow by adding a real company name to Airtable. Now our scenario is dynamic—our voice assistant can pitch intelligently based on lead-specific information.

In the next and final build, Build #3, we'll take this further: using all that call and research data to automatically generate custom proposals with PandaDoc, using OpenAI to write the content and e-sign links for instant client approval. By the end, we'll have automated the entire journey—from first contact to signed proposal—without a human lifting a finger.

1:22:09 – Build #3

Once a lead has expressed interest, it's the perfect moment to convert that momentum into a tailored business proposal. We already have the necessary context, so instead of waiting for manual work, we'll use OpenAI and PandaDoc to generate, send, and log a custom proposal automatically. In this final section, we'll add a proposal generator to the end of our workflow: OpenAI will create the custom text, and PandaDoc will assemble and deliver the proposal.

We'll use PandaDoc to create and send proposals with templates that include dynamic placeholders. Log in to PandaDoc (create an account if needed), go to Templates, and click + Template. You can start from an existing proposal template or use one you've already prepared. Inside a PandaDoc template, add tokens—placeholders replaced by real values (e.g., client company name). Example layout: a client introduction addressing them by name; a Goals & Plan section for the lead's priorities, proposed services, and an implementation plan. In the sidebar, add variables: proposal.goals (a paragraph summarising top priorities), proposal.services (a bulleted list of recommended services), proposal.implementation (a concise execution plan). In Pricing, include proposal.pricing (service/cost breakdown) and proposal.total (estimated total). Add signature and lead info in the agreement section. Name and save the template; you'll use it from Make's PandaDoc module. Optionally, style with logo and brand colours.

Back in Make, add PandaDoc → Create a Document. Connect your PandaDoc account, name the document based on the company, select the proposal template, set the recipient email, and map client info (company, first and last name). We'll generate the proposal token values via AI next. Enable Send document so PandaDoc emails the lead; add a subject and short message; save. Add OpenAI → Create a chat completion to draft clear, convincing proposal content. Choose a fast model and prompt it as a sales expert: include your service context (AI automation, agent-based systems), the client/company research (from earlier), and the voice-call summary. Instruct the model to identify the most relevant services and return JSON for easy mapping.

Test the OpenAI module with dummy call summary and research data. It should return well-structured JSON. Add Parse JSON to convert the OpenAI output into discrete variables for injection into PandaDoc. To test quickly, copy the expected JSON format from OpenAI and paste it as sample data into the Parse JSON module. With parsing confirmed, map the parsed fields to the PandaDoc token slots. Finally, update the lead's Airtable record so the sales team sees when the proposal was sent. In Proposal Sent On, write the current timestamp (now). This gives a clear trail: lead created → qualified → called (with summary) → interested → proposal sent.

Do a final scenario run to verify all moving parts. In Airtable, Proposal Sent On updates. Check the recipient's inbox: the PandaDoc proposal arrives, opens correctly, and reflects the client's needs and your plan. The client can sign and finish. PandaDoc also emails you when a lead views or completes a proposal, so you don't need extra Make steps to track that. You now have a full system that identifies and qualifies leads and translates interest into action.

Notes: The fixed delay after calls works in most cases, but if a call exceeds five minutes, the system might misclassify it as not answered. A robust solution is a webhook listener that waits for the call to end—outside this beginner scope. You could also extend the workflow to detect a signed proposal and trigger an onboarding sequence. Troubleshooting tips follow.

1:30:39 – Troubleshooting & Tips

You will encounter issues—platforms change, tutorials age, and odd bugs appear. That's normal. Even experienced builders spend significant time troubleshooting. Treat problem-solving as a muscle you strengthen. Use AI for support: describe the issue with context, screenshots, and error codes; let it search for current info. For complex problems, use deeper research; iterate with back-and-forth detail. Ask why a fix works to build understanding. Search Google/YouTube for recent threads and walkthroughs.

Tools like Google AI Studio can observe your screen and guide you in real time with voice. Example: diagnosing a 400 Bad Request due to JSON syntax and fixing array/string formatting. Learn to read documentation; use AI to translate docs into task-specific steps. Join communities (Discord, forums) for shared solutions. Success requires adaptability. The most successful persist, viewing obstacles as puzzles. When stuck, pause, then work your troubleshooting toolkit. Each solution compounds your future capability. Next, how to sell your systems to real clients.

1:36:55 – How to Sell AI Automations

You don't need to build "the next ChatGPT" to earn in AI. The real opportunity is helping businesses understand and implement automations like the ones you built. If you've reached this point, you're close to serving a market that's starving for help. As Kevin O'Leary notes, AI growth is exponential; the simplest path is helping people use the tech—implementation and execution—especially for small businesses. There are ~1.7M U.S. businesses making $500k–$10M annually; they create 62% of jobs, know they need AI, and lack time/skill to implement it. They need education, consulting, and implementation. Supply is scarce; demand is massive.

I started Morningside AI in early 2023 and have generated over $5M selling AI products and services. You don't have to be a technical genius—just one step ahead. Three service types: Education—workshops, staff training, courses. Consulting—map processes, identify high-ROI automations. Implementation—build and deploy automations. You can offer all three over time; you don't need to start with all. Your knowledge gap is your value. This video gives you a foundation; expand the gap to monetise.

Step 1: finish this video and build. Step 2: build more automations (free course on Skool) to deepen platform familiarity. After that, decide: go deeper technically for implementation, or start monetising via education/consulting. Be honest about strengths. If you love building, keep going—do the free course builds, then personal and friends' projects; within 2–3 months, you can sell implementation. If you don't love building, finish the free course and start monetising via education/consulting—you already have a usable knowledge gap.

Monetising doesn't always mean cold outreach. As Morgan Housel says, "Become the person someone wants to email about an opportunity before they call someone else." That's what I did. No cold outreach—people came via my content. Building in public (YouTube, LinkedIn) is the best advertisement. Share learnings and automations; this attracts clients. Start by reaching out to your network—friends, family, former colleagues. Send 5–10 messages explaining what you do, asking if they know anyone who needs help or if they can share. Warm outreach > cold outreach. Next, launch a posting campaign. Commit to 3–6 months of writing and video about AI automations: how they work, why they matter, your journey. I built my career on this; it's the highest-leverage move. Be consistent. Your goal isn't viral posts—it's to become the first person someone thinks of when they need AI help. Once you have clients, overdeliver and ask for referrals; that flywheel self-perpetuates. The opportunity is unprecedented; your entry timing is perfect. You've learned the fundamentals and built real systems. Now refine, deploy, and sell them. Let's wrap up.

1:44:29 – Conclusion

You made it to the end—I'm proud of you. This is a marathon video, and the fact that you're still watching means you're serious. That's the first and most important qualifier for success in AI. We've covered foundational concepts, built voice agents and lead systems, and explored monetisation. If you want to go deeper, join my free community on Skool. We have over 180,000 members, advanced courses, live sessions, and a Discord with dedicated support. You'll find the link in the description.

AI won't replace you—but a person using AI will. You've learned a powerful skill; now go build something valuable with it. Thank you for your time and trust. I'll see you in the next video.