My EO Master Series #3: AI Agents and Avatars
- David Hajdu
- Oct 13
- 5 min read
Beyond ChatGPT: Why 95% of Businesses Are Getting Zero ROI From AI (And How to Join the 5%)
Last week at Silk Lounge, our My EO Master Series brought together three founders who put AI first and are doing their best to crack the code on AI implementation.
In true EO spirit where members share their expertise rather than give advice, Dave Hajdu, David Nilssen, and Harley Trung opened up about their real-world AI experiments, failures, and wins.
The room was packed with familiar faces and newcomers, all wrestling with the same question: how do we actually make money from AI?
The answer surprised everyone.
The Uncomfortable Truth
A recent MIT study revealed that 95% of businesses spending billions on AI are getting exactly zero return on investment. Zero. Not because the technology doesn't work, but because most companies are stuck in the experimentation phase, using ChatGPT to be "more productive" without actually reducing costs or increasing revenue.
Think about your own company. Everyone's using ChatGPT, right?
But do you have a formal AI program?
Does someone lead it?
If you're nodding no, you're in the majority.
And you're leaving money on the table.
What Is an AI Agent, Really?
Before diving into use cases, the three founders shared their perspectives on what separates an AI agent from just using ChatGPT:
Harley Trung calls it "the next generation of Copilot, built to be autonomous, proactive, and adaptive." If Copilot is like having a co-pilot where you're still hands-on with controls, an agent is more like autopilot. You delegate, you manage, but you review results instead of doing the work. The key difference: you can deploy multiple agents simultaneously.
David Nilssen keeps it practical: agents complete multilayer tasks semi-autonomously. Instead of prompting ChatGPT over and over for each step, you stitch together multiple tasks into one automated workflow that runs on your behalf without constant input.
Dave Hajdu frames it simply: it's a really smart coworker that gets stuff done 24/7. But here's the critical distinction: it can only do one narrow task really well. You might need one agent to read emails, another to respond to them. They're specialists, not generalists.
The Military's Secret Weapon
Want to know who's crushing it with AI? The US military. Not because they have better technology, but because they have something most businesses lack: clearly defined Standard Operating Procedures.
Here's the insight that changes everything:
AI isn't your problem. Your undefined workflows are.
Before you build a single AI agent, map your processes. Identify what you do more than three times. Document your workflows.
Because AI's superpower isn't replacing work, it's systematizing and scaling what you already do well.
What Actually Works: Their Killer Use Cases
David Nilssen's businsses ($60M annual revenue) uses AI for lead enrichment that would've cost thousands in manual labor. His team uploads prospect lists to ChatGPT, which automatically searches the web, finds LinkedIn profiles, and creates personalized communication strategies for each target. But his favorite use of his digital twin? Sending company announcements as AI-generated videos in four languages to teams across the Philippines, Vietnam, Kenya, and Colombia. He only speaks two languages himself.
Harley Trung's software company has gone all-in on AI agents as team members. His developers work alongside AI agents that write code, review pull requests, and even conduct initial candidate screenings. But here's his secret weapon: in every one-on-one meeting, AI is the "third person in the room." He and his team members sit together with a computer, using AI live to give feedback, generate ideas, and solve problems collaboratively. It's not about using AI alone, it's about learning together.
Dave Hajdu's approach bridges the technical and business worlds. With 36 staff members and over 50 AI agents in his company, he's built what he calls a "Chief HR Officer" and "Chief Revenue Officer" (AI agents that handle everything from employee questions to sales research and contract generation). His newest venture, an AI education platform, uses AI avatars to teach courses in multiple languages. The kicker? He hates recording videos, so the AI does it for him.
The real breakthrough?
Understanding that AI agents aren't chatbots. They're 24/7 operations assistants that execute your processes, not just answer questions. Think of them as coworkers who never sleep, speak every language, and cost a fraction of a salary.
The Two Things Killing Your AI ROI
Based on the workshop discussion, businesses fail at AI for two reasons:
First, disorganized data. Your customer feedback, brand guidelines, workflows, if they're not documented and accessible, AI can't help you. Garbage in, garbage out.
Second, no leadership. Someone needs to own your AI strategy. Not just use it casually, but systematically identify opportunities, train your team, and measure results.
How They're Training Their Teams
Getting your team to actually use AI effectively requires leadership and structure. Here's how each founder approaches it:
David Nilssen created a formal AI program with four components: an AI community of five people overseeing implementation, an internal community where AI enthusiasts share wins and challenges, weekly 15-minute micro-trainings partnered with the AI Officer Institute, and quarterly innovation awards (cash grants for team members who demonstrate 10x ROI on AI deployments).
Harley Trung brings the office back for a reason. He puts a computer between himself and team members during one-on-ones, making AI the "third person" in every meeting. They look at reports together, use AI together, and learn by doing. His rule: teach yourself first. When you teach someone else how to use AI, you learn it better yourself.
Dave Hajdu requires each team member to produce one AI learning per month. His operations lead reviews all submissions, identifies patterns where multiple people are learning the same thing, and creates "micro sessions" to train the whole team. Simple incentive: it's one of three requirements for raises (improve yourself, hit your goals, and help the team perform).
Building Your Digital Twin
The workshop's hands-on component focused on HeyGen, a tool for creating AI avatars that speak in any language, a key part of how these founders scale their personal presence.
David N sends company-wide announcements as videos in four languages: Filipino, Vietnamese, Spanish, and English. He records once, and his digital twin delivers personalized messages to every team across continents.
Dave's using his avatar to teach his online AI courses, eliminating the need to record dozens of videos himself. The avatar speaks with his mannerisms, in multiple languages, handling the detail-heavy content while he focuses on live overview sessions.
The process is surprisingly simple: record a two-minute video of yourself talking naturally (keep your head still, focus on the camera), upload to HeyGen, and it creates your avatar. From there, you can generate videos by simply pasting in scripts. The avatar handles the rest, including automatic translation.
The group walked through creating customer response videos for Silk Lounge in English and Korean, demonstrating how a local business could provide personalized, multilingual customer service at scale.
The key insight: your digital twin amplifies you, it doesn't replace you. It handles the repetitive, scalable communication while you focus on high-value interactions.
The Bottom Line
By 2026, every employee will likely have an AI agent as a coworker. The question isn't whether you'll adopt AI, it's whether you'll lead the charge or scramble to catch up.
Technology always catches up.
Your choice is whether you want to lead or lag.
Want to dive deeper? Join us at the next EO Vietnam Master Series where founders share what's actually working in their businesses.
Scroll down to get the resources we shared during the session 👇
Session Materials
📑 Recorded Session and Slides:
🧠 Exercises:
Build Your Workflow → https://qrco.de/afd_buildworkflow
Create Your Digital Twin → https://qrco.de/mydigitaltwin
💬 Continue the learning:
Join the Micro Session here → https://community.ai-officer.com/c/ai-labs-micro-sessions/ai-labs-building-your-digital-twin
Sign up and try Hey Gen
Thanks again for being part of this learning journey, see you in the next session!
Your Fellow EO Members,
Dave Hajdu, David Nilssen & Harley Trung