Inside Intercom’s AI Turnaround, Why Software Companies Must Train Their Own Models | Eoghan McCabe, Co-founder and CEO
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
Owen McCabe, co-founder and CEO of Intercom and Fin, shares how his company became one of the first late-stage software companies to successfully pivot to become AI-native. After leaving the CEO role in 2020 due to illness and returning in 2022 to a struggling company with five quarters of declining net new ARR, McCabe made the bold decision to bet the entire company on AI customer service. (13:58) The month after his return, ChatGPT launched, creating the perfect storm for transformation. Today, Fin processes over one million customer service resolutions per week and wins 100% of competitive bake-offs against direct competitors. (15:36) McCabe discusses the future of AI agents, outcome-based pricing models, and why he believes AI will eventually be better than humans at most white-collar work.
- Main themes include AI transformation of traditional software companies, the superiority of AI over humans in customer service, and the challenges of building AI-native products in a hyper-competitive landscape
Speakers
Owen McCabe
Co-founder and CEO of Intercom and Fin, Owen McCabe started the company in 2011 and led it through multiple phases of growth. After stepping down as CEO in 2020 due to illness, he returned in 2022 to orchestrate one of the most dramatic AI pivots in late-stage software. Under his leadership, Intercom became the largest service AI agent company by revenue and customer count, winning 100% of competitive bake-offs. McCabe is originally from Dublin, Ireland and has been a vocal advocate for authentic founder communication and direct customer engagement.
Turner Novak
Host of The PEL podcast and founder of Banana Capital, Turner Novak explores the world's greatest startup stories through in-depth conversations with founders and leaders. He focuses on understanding the strategic decisions and pivotal moments that define successful companies, particularly in the technology and venture capital space.
Key Takeaways
Bet the Company When You Have Nothing to Lose
McCabe returned to a struggling Intercom with five quarters of declining net new ARR and the moral authority to make radical changes. (13:21) This precarious position actually became an advantage - when you're already in trouble, you have the freedom to make bold bets that companies with "more to lose" cannot make. He emphasizes that many successful pivots come from necessity rather than choice, and having your back against the wall can paradoxically create the conditions for breakthrough innovation.
AI Will Be Better Than Humans at Most Service Work
McCabe demonstrates this with concrete data: human service reps take an average of 9.8 minutes to close a ticket and make customers wait 102.8 minutes before responding, while Fin provides essentially instant responses. (18:42) Beyond speed, AI agents maintain consistency, never make mistakes due to fatigue, and follow procedures perfectly. This isn't just about efficiency - it's about delivering objectively better service through "better, faster, cheaper - pick all three" rather than the traditional "pick two" limitation.
Own the Full Stack to Win in AI
Companies that rely solely on generic models are "subscribing to eventually becoming generic," McCabe explains. (25:57) Intercom has built its own AI models, uses dozens of different models from various vendors, and employs over 50 AI researchers and engineers. This approach allows them to maintain healthy margins at 99¢ per resolution while competitors charge $2-4, and consistently outperform in head-to-head comparisons. The Apple analogy is apt - owning hardware, OS, chips, and software creates competitive advantages that pure software players cannot match.
Outcome-Based Pricing Aligns Incentives Perfectly
Instead of charging for "work done," Fin charges 99¢ only when it successfully resolves a customer issue to their satisfaction. (43:56) If the AI fails or passes the issue to a human, there's no charge. This model eliminates "dead revenue" and creates a perpetual incentive for the company to improve resolution rates. McCabe contrasts this with traditional pricing where companies charge $26 per human resolution regardless of outcome, making the value proposition impossible to argue against.
Brand Separation is Critical for AI Product Success
McCabe created Fin as a separate brand rather than "Intercom AI" because established brands can become "a ball and chain" when new technologies emerge. (70:48) People unconsciously expect AI innovation from young, new companies, not established players. He spends tens of millions on advertising with no Intercom branding, sending traffic to fin.ai, with Intercom mentioned only at the bottom. This strategy recognizes that even respected brands must earn trust anew in fundamentally different technological paradigms.
Statistics & Facts
- Intercom had five quarters of sequentially declining net new ARR before McCabe's return, demonstrating how quickly late-stage software companies can deteriorate. (12:31)
- Fin processes over one million customer service resolutions per week, which McCabe calculates as equivalent to replacing 6,533 human workers. (23:37)
- Human customer service representatives take an average of 9.8 minutes to close a ticket and make customers wait 102.8 minutes before responding, compared to Fin's essentially instant response times. (18:42)
- Intercom charges $26 per human resolution internally for their own customer service, highlighting the dramatic cost advantage of their 99¢ AI resolution pricing. (44:19)
- McCabe's team has conducted over 400 A/B tests to optimize Fin's performance, demonstrating the scientific approach required for AI product development. (51:11)
