Back From the Brink: How Intercom's Founder Rebuilt a Stalled SaaS Company Around the Fin AI Agent

McCabe opens with the current numbers. Benchmarked against the roughly 120 public B2B software companies, Intercom sits at about the 15th percentile for ARR growth. Fin, its AI agent, is growing north of 300%, went from one to 12 million ARR in its first year, and will pass 100 million ARR in less than three quarters. He says Fin leads the customer experience category by customer count and revenue and wins head-to-head bake-offs.

The starting point was grim. McCabe left the CEO role in 2020 after years of illness, and the company became bloated, unfocused and "trying to do all the things for all the people." Growth fell to low single digits, and five straight quarters of falling net new ARR meant it was about to hit zero, which would have meant negative growth. He returned to stop it, and ChatGPT was announced a month later.

Intercom happened to have an AI group already, building rudimentary bots and its own machine-learning Q&A, and those engineers saw GPT-3.5 as different. About six weeks after the launch the team had a beta of what became Fin, and the company had a large base of 30,000 paying customers and billions of data points to work with. McCabe says he is happy to take credit for being brave, but notes they were already in a tough spot with no choice.

Lenny notes the irony that Intercom was once skeptical of bots as impersonal. McCabe agrees that the early mission was to make internet business personal, and says he asked himself whether that still made sense. His answer is that you cannot fight the future and must be part of it.

He has come to believe that a charismatic, consistent, instantly available expert for every customer at all hours is more personal than making them wait two to four days for a canned response. He admits it could be post-rationalization but says that is where he stands after examining it.

To illustrate, he points to Waymo, which he says crashes 3.5 times less than human drivers and avoids the annoyances of a talkative ride. He loves humans and wants them in the rest of his life, but for the practical glue between the human parts of life he would rather have AI and robotics.

Lenny raises Intercom's reputation for confusing and expensive pricing, and McCabe confirms that it was the subject of viral memes. He blames two things: an unfocused strategy, which led to metrics in every direction (seats, messages, leads) and tiers and gates for many customer sizes, and an unwillingness to make bold decisions and take short-term pain.

When he came back he simplified the pricing even though it meant lowering prices for many customers. He says Intercom has already given away something like $50 million in ARR this way. Simpler, more predictable and fairer pricing keeps customers longer, and it also reinforced the value of being customer-obsessed inside the company.

Lenny points out that the company was worth multiple billions and earning hundreds of millions, so going all in was still a risk. McCabe says several things combined: public software stocks had dropped 80 to 90%, Intercom's growth had cratered to low single digits, and the founders were impatient to make something of the company.

He also cites his own anger about how the company was run and the compromises he had made to placate employees and investors by following industry best practices. He says founders betray their intuition in small pieces over the years. So he took what he calls an authoritarian, top-down, founder-first approach, which he found cathartic, and the AI opportunity matched his intuition.

He warns that founders tend to rewrite such stories to be tidy and flattering, and says the real story was a messy cocktail of logic, desperation and gut feeling.

Lenny cites a statistic that early on Fin cost more than it charged, roughly 120 cents per dollar of revenue, and McCabe confirms it. Intercom charges 99 cents when Fin resolves a customer's problem, so the revenue is aligned with the metric by which agents are judged. He says the company is proud of having the highest resolution rate and obsesses over it.

The price came from research showing many companies spend $20 to $30 per human-resolved ticket, followed by testing how low to go. McCabe says early on they sensed customers would not value digital work as much as human work, even though it is better and always available. His view is that if someone will not pay 99 cents to have their customer's problem solved well, there is no business.

He insists pricing should come from value, not cost: cost is the company's problem, and Intercom had the intuition that it would fall, which it did. Lenny says the approach matches pricing expert Madhavan's advice on simple, outcome-based pricing, and McCabe is relieved that Intercom's pricing wars are over.

McCabe lists what he did. He cut costs aggressively and stopped projects, including fitting out a large new office. He then chose to focus on customer service, noting that Zendesk had recently been acquired and, in his view, was strategically and culturally dead, even though staff argued the company still had about $80 million of ARR in other products.

He made that decision unilaterally instead of running the usual professional-CEO process of collecting input in a spreadsheet, because nobody was making decisions. When AI arrived he announced that Intercom would spend nearly $100 million of its own cash on it. He also rewrote the company values as a "sharp knife" emphasizing resilience, high standards, hard work and shareholder value.

Quarterly reviews gave each person a grade for goals and a score for behavior against the values, using a formula he coded himself so managers were out of the loop. People below a threshold were thanked and let go. He says there was pain, attempts at a soft coup and letters to the board, but a survey 15 or 16 months in showed 98 to 99% approval of management, leadership and the new strategy.

He sums up the story as strategy, go-to-market and pricing, an AI bet, and culture, adding that none of it would have mattered without the AI bet: "we built Fin and that changed everything."

Asked about the soft coup, McCabe says that changing the rules for people who had been told they were in control creates friction. He believes great companies have a clear hierarchy in which the CEO makes brave, hard decisions, uses experts as input and takes responsibility for the outcome, including being fired if they fail. He adds that founder-led companies tend to outperform because founders have the authority and appetite for risk.

Turnover was ultimately around 40% over a couple of years. Culture started changing within a quarter because it is set by a small number of people, but hiring people with the new level of ambition took longer. He talks about "employee, founder, product market fit" and says people who prefer a gentler, democratic environment will be happier elsewhere, so a push out the door is doing them a favor.

He never regretted coming back, since leaving had felt like betraying his own dream, but he often does not enjoy the job. Most days are bonus policies, comp proposals and eight to ten meetings, not the swashbuckling adventure of his imagination. He stays because of the broader mission.

McCabe stresses that customer experience (service, success, sales and marketing) is deceptively large, usually the biggest part of a company's headcount. He expects any function with repetitive operational work, such as invoicing or onboarding, to be automated, with a mix of generic operations bots and expert agents for fields like law and accounting.

He pictures organizations as a medley of humans and agents rather than humans on top and agents below, with people in individual-contributor roles configuring, monitoring and covering edge cases. Organizations will be smaller and flatter, he thinks, and he can imagine an agent chief of staff. He expects this to be deflationary, with more competition and cheaper options for consumers.

On jobs, McCabe says technology has always removed repetitive, demeaning or dangerous work while population, GDP and longevity grew, and he will not apologize for competing for such work. He expects CX roles, basic repetitive roles and parts of sales to shrink, which puts him slightly at odds with Marc Benioff, though he agrees that human connection and trust in sales will remain.

Lenny mentions that Anthropic co-founder Ben Mann said even his own job would be automated. McCabe responds that efficiency is not the top thing people value, and that as AI makes automated things nearly free, people will pay more for beauty, craft, human stories and imperfect individuals.

McCabe's core advice, which his co-founder Des is writing a book about, is that you do not have a choice. He thinks AI is bigger than microprocessors, PCs, the internet or mobile, and will disrupt all categories in the most aggressive ways. He urges leaders to roll up their sleeves, find out what will disrupt them, and bring in real AI scientists and leaders, naming Fergal Reid, who is about to be promoted to chief AI officer.

He says young AI companies are working 12 hours a day, every day of the year, and that this is the price of entry. If an incumbent's founder is unwilling to do that, he says, they should hire a younger person to run it. He adds that younger companies use AI by default for coding, creative work and even job descriptions, while most people at companies of his generation do not.

Lenny recalls the Perplexity founders asking ChatGPT before asking a teammate. McCabe says the culture clash between generations is real and the sooner leaders understand it the better. He adds that he does not generally promote working this hard, believes a good life includes slow walks in nature, and is only describing the cost of competing in AI.

Lenny comments on how reflective and centered McCabe is and asks whether he changed. McCabe names three factors: 14 years of startup life, which either kills you or makes you stronger; about 12 years of therapy and spiritual work with a coach and therapist named Yosi Amram; and the two years away, when he was sick and attacked in the press and his ego was "eviscerated."

He argues that good therapy takes away the counterproductive edges, such as insecurity, jealousy and getting triggered, helping you acknowledge what you are not good at and communicate more authentically. The image of yourself as a perfect, brilliant leader, which successful founders tend to form, had to die, because it makes any challenge feel threatening.

When Lenny shows him a piece of art reading "ego death now," McCabe says the ego never fully dies and cites Ram Dass saying the edges simply got smoothed. Acknowledging and even loving your ego, he says, means you can notice jealousy and laugh at it.

Lenny shares research showing Intercom among the companies whose alumni most often become CPOs, early PMs and founders, and asks why. McCabe says the culture was product-centered, driven by him and Des Traynor as product people, and that good people wanted to work there. The sprawling strategy required many PM groups with a lot of autonomy, so PMs acted like mini CEOs.

He adds a first-principles approach to everything, with frameworks for any question, such as who an event is for, what its goal is, and how success is defined. He points to the chief product officer Paul Adams's new book on AI and customer service and to the company's love of whiteboards. Many alumni became founders because Intercom hired founder types and told them to learn how great companies are built there and then start their own.

Lenny says the themes of complexity, ownership and first-principles thinking recur across other companies on the list, many of them fintechs, and notes that Intercom popularized frameworks like RICE and jobs to be done. McCabe closes with his reason for appearing: to make people see Intercom as a startup-like late-stage company, and to point them to Fin at fin.ai.