I've spent the last year running an AI bootcamp for executives. Not a lecture series. Not a webinar where someone demos ChatGPT. A three-session sprint where CEOs, CTOs, and VPs sit down with Claude Code and build working AI agents for their businesses.
Twenty executives have gone through it so far. Every one of them came in skeptical about at least one thing. Some thought they didn't have the technical background. Some thought AI agents were hype. Some thought they just needed to hire someone to handle this.
Every one of them left with production tools running in their business.
Here's what they built, what surprised them, and what I learned about how executives actually adopt AI.
The Bootcamp
Quick context on the format. It's three sessions, compressed over one to two weeks. Not a lecture. Not slides. Building.
Session 1 is foundation. We cover the mental model: AI workers, not AI tools. The idea that you're building a workforce of agents, each with a defined job, defined inputs, defined outputs, and quality checks. By the end of Session 1, every participant has built their first working agent.
Session 2 is the method. Agent orchestration. How to chain agents together. How to build verification into every workflow. How to design for failure modes. The second session is where things click. People stop thinking about AI as a novelty and start thinking about it as infrastructure.
Session 3 is go live. We take everything from Sessions 1 and 2 and build a custom setup for their business. By the end, they have a working system with multiple agents handling real work.
The price is $5,000. That's deliberately accessible. I wanted to remove the "let me think about it" barrier and get people building.
Now let me tell you about the builders.
The CFO Who Automated Finance Ops
Sarah (not her real name) is the CFO of a 120-person SaaS company. She came to the bootcamp because her CEO told her to. Her words: "I figured I'd sit through it and then hire someone to actually do whatever this is."
By the end of Session 1, she'd built a Financial Reporter agent. It pulls data from QuickBooks and Stripe, reconciles the numbers, and generates a weekly P&L summary with narrative commentary on variances. The thing that used to take her team four hours every Monday morning now runs automatically Sunday night.
But here's where it gets interesting. In Session 2, she built a Cash Flow Planner. It takes the financial data, combines it with pipeline data from their CRM, and projects 90-day cash positions under three scenarios: base case, optimistic, and conservative. She'd been doing this manually in a spreadsheet every month. The agent does it daily.
In Session 3, she built an Expense Analyzer that flags unusual spending patterns across departments and a Board Prep Assistant that compiles the metrics and narratives she needs for quarterly board meetings.
Four agents. Built in two weeks. Collectively saving her team roughly 15 hours a week.
The thing Sarah said that stuck with me: "I kept thinking I needed to understand machine learning to do this. I don't. I need to understand my workflows. I already knew those."
That's the insight most people miss. Building AI agents isn't about understanding AI. It's about understanding your work well enough to describe it precisely. Executives are uniquely good at this because they already think in terms of processes, inputs, and outputs. They just didn't realize that's all an AI agent needs.
The CX Lead Who Built a Command Center
Marcus runs customer experience for a 200-person B2B company. His team handles 400+ tickets a week across email, chat, and phone. He was drowning in triage, escalation, and reporting.
He came to the bootcamp looking for a chatbot. He left with a CX Command Center.
Here's what he built:
A Triage Agent that classifies incoming tickets by urgency, category, and required expertise. It reads the ticket, pulls the customer's history, checks for similar resolved issues, and either drafts a response for simple cases or routes complex ones to the right specialist with a pre-built context packet.
A Pattern Detector that monitors ticket volume and content daily, flagging spikes and emerging issues. "We used to find out about product bugs from our customers. Now we know before the third ticket comes in."
A Weekly CX Report agent that compiles resolution times, satisfaction scores, ticket volume by category, and trend analysis. He used to spend Friday afternoons building this report manually.
The shift in his team's work was immediate. His agents went from "respond to tickets" to "manage the response system." Same headcount. Completely different altitude. They spend their time improving response quality and handling the genuinely complex cases that need human judgment.
Marcus told me something I've heard in some form from almost every bootcamp graduate: "I didn't know this was possible without a full engineering team."
He was right that it wasn't possible a year ago. The tools have caught up to the point where an ops leader with clear workflow definitions can build production agents without writing traditional code.
The Creative Director Who Got Her Weekends Back
This one surprised me. Alex is a creative director at a mid-size agency. She came to the bootcamp at the recommendation of a friend, not because she thought of herself as technical.
She built a Social Media Manager agent.
The workflow: Alex spends 15 minutes on Monday morning recording a voice memo about the week's priorities, themes, and any specific things she wants to talk about. The agent takes that briefing, generates a full week of social content across three platforms, formats it for each platform's requirements, and stages it for review.
Before the agent, she spent six to eight hours a week on social content. Now it's 15 minutes of input plus 30 minutes of review and editing. She described it as "getting my weekends back" because social content was the thing that always bled into Saturday.
But the real win came in Session 3, when she built a second agent: a Client Brief Analyzer. It takes incoming client briefs, extracts requirements, compares them against the agency's capability matrix, and generates a preliminary scope estimate with flagged risks. Something that used to take two hours of reading and a meeting now takes the agent eight minutes.
Alex's experience illustrates something important about the bootcamp: it's not just for technical people. It's for anyone who has recurring workflows they can describe. The more precisely you can describe what you need done, the better the agent performs. And creative directors are, it turns out, extremely precise about workflow descriptions.
The Career Changer Who Built a Personal HQ
Not every participant is running a company. David is a mid-career professional who was exploring a move into AI-adjacent work. He came to the bootcamp to understand the technology well enough to be credible in interviews and conversations.
He built what I call a Personal HQ: a command center for his career transition.
The core agent is a Research Assistant. David feeds it a company name, and it compiles a dossier: recent news, funding history, leadership team backgrounds, tech stack (from job postings), competitive positioning, and potential conversation starters. He uses it before every networking conversation and interview.
He also built a Job Market Analyzer that monitors postings in his target roles, tracks which companies are hiring, and identifies patterns in required skills. And a Personal CRM agent that tracks his networking conversations, follow-up dates, and relationship status.
David didn't come to the bootcamp to build a business tool. He came to learn. But the methodology is the same whether you're automating company operations or your own life. An AI worker is an AI worker. The architecture patterns transfer.
Two months after the bootcamp, David landed a role as an AI Operations Lead at a Series B company. His interviewer told him that the fact he'd built working AI agents, not just studied them, was the deciding factor.
The Consultant Who Built a Framework
This one is my favorite, because it's the most meta.
Rachel is an independent consultant who advises restaurants and hospitality businesses on operations. She came to the bootcamp because she kept getting asked about AI by her clients and didn't have good answers.
She built a Restaurant Operations Framework: a set of AI agents that her clients can deploy.
The first agent handles inventory forecasting. It takes historical sales data, current bookings, seasonal patterns, and local events, then generates ordering recommendations. Her pilot client reduced food waste by roughly 20% in the first month.
The second agent is a Staff Scheduling Optimizer that takes projected covers, employee availability, and labor cost targets, then generates draft schedules. Not a replacement for the manager's judgment, but a starting point that accounts for all the variables.
The third is a Customer Feedback Analyzer that processes reviews across Google, Yelp, and internal comment cards, identifies recurring themes, and generates a weekly briefing for the management team.
Rachel didn't just build tools for herself. She built a productized offering. She now sells the framework as part of her consulting engagements. The bootcamp cost her $5,000. Her first implementation engagement was $15,000. The ROI math worked out.
Patterns Across 20 Builders
After watching 20 executives go through this process, here's what I've learned:
Worker 1 Is the Hardest. Worker 10 Is Almost Templated.
Every participant struggles with the first agent. Not because it's technically hard, but because the mental model is new. You have to learn to think about your work as a set of defined tasks with specific inputs and outputs. That's an unfamiliar framing for most people.
But once the first agent works, something shifts. You start seeing agents everywhere. "Could an agent do this?" becomes a reflexive question. The second agent is faster. The third uses patterns from the first two. By the fifth, you're composing agents and connecting outputs. By the tenth, you're stamping them out from templates.
The compound effect is real and consistent across every participant I've worked with.
The Best Agents Come from Deep Domain Knowledge
The agents that work best aren't the ones built by the most technical people. They're the ones built by people who deeply understand their workflows.
Sarah's financial agents work because she knows exactly what a good P&L summary looks like and exactly where the data lives. Marcus's triage agent works because he's personally handled thousands of tickets and knows the classification patterns. Rachel's inventory agent works because she's spent years watching restaurants order wrong.
AI agents are an amplifier for domain expertise. The people who know their business cold build the most effective agents, regardless of their technical background.
Executives Are Better at This Than Engineers
This sounds counterintuitive, but I've seen it consistently. Engineers tend to overthink the architecture. They want to build robust systems with error handling and retry logic before the first version works. Executives think in terms of "what do I need and what does done look like?" That's exactly the right framing for building AI agents.
The best agent builders are people who can clearly describe: here's the input, here's what I want the agent to do with it, here's what good output looks like, and here's how I'll know if it's wrong. That's executive thinking, not engineering thinking.
The Transfer Is Permanent
This isn't like a workshop where you learn something and forget it in a month. The agents keep running. The methodology becomes how you think about work. Six months after the bootcamp, every participant I've followed up with is still building new agents.
Most have built two to three times the number of agents they built during the bootcamp itself. The methodology sticks because it delivers visible, immediate value. You don't need discipline to maintain a habit that saves you ten hours a week.
What This Means
Here's the thing I keep coming back to: these are not technical people. They're a CFO, a CX lead, a creative director, a career changer, and a restaurant consultant. None of them wrote code before the bootcamp. None of them would describe themselves as "technical."
But every one of them built production AI agents in two weeks. And every one of them is still building.
The gap in AI adoption isn't a technology gap. It's a methodology gap. The tools exist. The models are capable. What's missing is a structured way for non-technical operators to build with them.
That's what the bootcamp provides. Not AI education. Not AI strategy. A method for building AI workers. Once someone has the method, they don't need me anymore. They just need their own domain knowledge and the time to build.
77% of employers plan to reskill their workforce for AI, but only a third of employees have received any AI training. The training that exists is mostly lectures and demos. Nobody's asking executives to build.
We are.
The next cohort of the Exec AGI Bootcamp starts soon. If you're an executive at a growth-stage company and you want to understand AI by building with it, not by watching someone else demo it, this is the fastest path I know.
Details and booking at advisory.getindigo.ai. Three sessions. $5,000. You'll leave with working AI agents and the methodology to build more.
