1508: This Founder Runs His Entire Company on an Army of AI Agents w/ Drew Mehta
Tired of hearing about AI as a vague, futuristic concept? In this episode, we pull back the curtain on what it really looks like to build and run a business on a fleet of AI agents. Drew Mehta, founder of ThinkFraction, joins us to share how he went from managing products for Major League Baseball and a $750B asset management firm to launching his own AI-native company. He doesn't just talk the talk; his company’s content, outreach, and research are all powered by AI agents he designed, with him as the "human in the loop." This episode is a masterclass in strategic AI implementation, teaching you how to stop chasing buzzwords and start solving real, costly business problems. You'll learn how to identify the right opportunities for automation, why human judgment is more valuable than ever, and how to gain a massive competitive edge without being a tech genius.
💡 What You'll Take Away For YOUR Business
- How to Build Your Own AI Team: Learn the structure behind Drew’s "fleet of agents" that ideates content, drafts posts, and runs research, so you can replicate it in your business.
- The #1 Mistake Businesses Make with AI: Discover why starting with the problem, not the tool, is the key to a positive ROI and avoiding costly over-engineering.
- The "Intern" Framework for AI Risk: A simple but powerful mental model for deciding which tasks are safe to fully automate and where you absolutely need a human in the loop.
- Deterministic vs. Probabilistic AI: Understand the crucial difference and when to use simple automation versus complex AI agents to save money and get better results.
- Turning Operations into an Advantage: Learn why your internal processes and bottlenecks are the biggest untapped opportunities for leveraging AI to become faster and leaner than your competition.
- Enhancing the Customer Experience with AI: See how AI can handle administrative tasks to free you up, improve response times, and create a better experience that wins you more customers.
📝 About Drew Mehta
Drew Mehta is the founder of ThinkFraction and a 20-year veteran product manager who has built solutions for some of the biggest names in sports and finance, including Major League Baseball and the global asset management firm KKR. A problem-solver by nature, Drew became fascinated with AI's potential to solve complex business challenges. Instead of just advising from the sidelines, he built his entire company, ThinkFraction, on a foundation of AI agents he directs. This unique, hands-on experience gives him a practical and grounded perspective on how any business can leverage AI to operate more efficiently and effectively.
🎯 Drew’s BEST Piece of Advice for Wantrepreneurs and Entrepreneurs
"The biggest thing that stops you now is yourself. So don't let that be the reason why you don't take a risk."
📢 Memorable Quotes
- “Everyone else grabs the tool and goes hunting for the problem, whereas I start with the problem.” – Drew Mehta
- "You don't want to bring a machine gun to a pillow fight... Point AI at something a simple rule could have handled and you're paying a premium to do the job worse.” – Drew Mehta
- "If you're not willing to let an intern do something, then I wouldn't let an AI do it either, especially with the stakes of your business." – Drew Mehta
💡 Actionable Takeaways
- Identify Your Biggest Time Sink: For one week, track your tasks and identify the single most repetitive, time-consuming operational activity in your business. This is your first candidate for AI automation.
- Apply the "Intern Test": Before automating a customer-facing or high-stakes process, ask, "Would I let a new intern handle this without my final approval?" Use the answer to decide where the "human in the loop" is non-negotiable.
- Focus on a Workflow Problem First: Don't try to change your core product or service with AI. Instead, look for a workflow bottleneck, like lead intake or invoicing, and solve that first for a quick, measurable win.
- Start with Simple Rules, Not Complex Agents: Before building an entire agentic system, see if a simple, rule-based automation (like an "if-then" workflow) can solve 80% of the problem.
🔗 Links & Resources
