How Intercom rebuilt itself as an AI-first business using Finn

Executive overview

Intercom was in serious trouble: five consecutive quarters of declining net new ARR, nearly hitting zero growth, and a bloated, unfocused strategy. Eoghan McCabe returned as CEO one month before ChatGPT launched and used the AI moment to force a complete reset — strategy, culture, pricing, and product.

The result is Finn, an AI customer service agent now approaching $100M ARR, growing 300%+. Intercom went from low single-digit growth to the 15th percentile of all public software companies in under three years.

When your back is against the wall, AI transformation is not a bet — it's the only move.

The state of the business before the reset

• Post-COVID revenue sugar rush masked a deteriorating core: five quarters of sequential decline in net new ARR
• Strategy was spread across too many products for too many customer segments
• McCabe had stepped back in 2020 due to illness; the company drifted toward comfortable, consensus-driven decision-making
• Culture had become bloated, complaint-heavy, and risk-averse

The pivot decision

• GPT 3.5 launched while McCabe was one month back as CEO; the AI team had a working Finn prototype six weeks later
• Decision to go all-in was driven by desperation, intuition, and founder impatience — not clean strategic logic
• Early unit economics were brutal: charging $0.99 per resolved ticket while cost per resolution was ~$120
• They committed ~$100M in internal cash to AI investment despite having no guarantee of model cost improvement

Strategic choices that made the shift work

• Dropped all other product lines and focused entirely on customer service (CX)
• Identified Zendesk's post-acquisition weakness as a market opening
• Simplified pricing: wrote off ~$50M in ARR to give customers simpler, fairer plans
• Adopted outcome-based pricing at $0.99 per resolved ticket — revenue tied directly to delivered value
• Finn is now the largest AI CX agent by customer count, revenue, and performance benchmarks

Culture reset and founder mode

• Rewrote company values to emphasize resilience, high standards, hard work, and shareholder focus
• Implemented quarterly performance reviews with hard-coded formulas: score on goals plus score on values behavior
• Low scorers were exited regardless of manager preference — removing human bias from the process
• Faced significant internal resistance: a letter to the board, attempted soft coup, widespread unhappiness
• ~40% employee turnover over two to three years
• 15–16 months after the reset began: 98–99% management approval in anonymous survey

Pricing philosophy

• Old pricing was a viral meme for complexity and opacity — a symptom of the unfocused strategy
• New principle: pricing should reflect value, not cost; cost is an internal problem to solve
• $0.99/resolution was chosen as the nexus of maximum customer palatability and business viability
• Customers were previously paying $20–30 per resolved human ticket; Finn dramatically undercuts that

The AI opportunity and what companies still get wrong

• CX is deceptively large: encompasses service, success, sales, and all customer engagement — the biggest headcount function in most businesses
• AI will also hit: invoicing, onboarding, offboarding, contract review, and any high-volume repetitive operational work
• Future organizations will be smaller, flatter, and include agents in management-adjacent roles
• Companies "doing AI" by sprinkling it on top of unchanged cultures will lose to startups running 12 hours a day, 365 days a year
• The compounding advantage of young AI companies is not just hours — it's default AI-native workflows at every step

On jobs and disruption

• Historical technology waves removed dangerous, demeaning, repetitive work while population and GDP grew
• AI will reduce headcount in CX and parts of sales (especially SDR/qualification roles)
• What survives: human connection, trust, judgment, creativity, and the value people place on human craft and beauty
• Abundance of AI-generated content makes human-authored, high-craft work more valuable, not less

What produces great product leaders (Intercom's track record)

• Product-first founding culture: strategy and design were always CEO-level concerns
• Sprawling strategy created many autonomous PM roles — effectively mini-CEO positions
• First-principles framework-building was taught and modeled from the top (RICE, jobs-to-be-done)
• Deliberately hired founder-types who wanted to learn company-building and move on
• Whiteboard culture: problems were always decomposed into mechanisms, goals, and definitions of success before solutions

Personal transformation and leadership

• 12 years of weekly therapy with a CEO-specialist coach (Yossi Amram) removed counterproductive edges without diminishing drive
• Two years away from Intercom — burned out, publicly attacked, revenue declining — destroyed an ego identity that had become limiting
• The result: clearer self-knowledge, less reactivity, more authentic communication with teams
• McCabe's view: ego never fully dies, but great therapy smooths the edges that make leaders ineffective
• Coming back was driven partly by not wanting to abandon something he'd built; departure had felt like a betrayal