Did we miss a partner program update from any of the major players which you feel is important?

This week Salesforce agreed to buy Intercom (now rebranded as Fin) for $3.6 billion. A 15-year-old SaaS workflow company "left for dead" managed to complete a tremendous turnaround and reinvent itself as an agentic AI business. There are useful lessons here for AI services operators, many of which are pivoting traditional dev shops or consulting firms into AI enablement. 

If you missed it, Intercom's CEO Eoghan McCabe wrote a viral essay in March about what it took to reinvent the company around AI. It's a great read, and the acquisition is a fitting capstone to that story.

The turnaround was remarkable by any measure. Intercom didn't just change the company's trajectory. It launched an entirely new product that put its existing business at risk, and did it anyway. That takes conviction that most management teams won't find.

If reinvention is possible for a 15-year-old, 1,000-FTE company, a nimble services business has every opportunity to undergo its own transformation.

We're already seeing early signs of this. AI GTM agencies are a good example: firms that were traditionally focused on CRM implementation now offer Clay services and build custom AI enabled workflow products connecting disparate tools. They didn't throw out what they knew, but they deprioritised it in the value chain and transformed their employee base to hire experts in AI GTM tools.

There are plenty of traditional mid-market training, consulting and development firms whose businesses are still going strong. But the writing is on the wall. Use this as proof that you can pivot and compete with AI-native vendors on equal footing. Your existing customer relationships are your biggest leverage. An AI-native startup would pay a great deal for what you already have.

Fin built Apex: a model trained on real-world customer service data from 14,000+ companies, achieving 76% resolution rates and outperforming both OpenAI and Anthropic on this specific task. That's a genuine data moat, earned through years of customer deployments.

Sierra, Fin's closest AI-native peer, took a different path entirely. They use 15+ foundation models and see their role primarily as an orchestration layer: routing each part of an enterprise interaction to whichever model handles it best. Only two components in their value chain are proprietary. Yet Sierra is valued at $15.8 billion: a higher multiple than Fin's parent company achieved in a full exit. Owning a model didn't close the gap.

Many founders we speak with - indoctrinated by years of SaaS mantras - believe they can't build a scalable, profitable, valuable business without shipping a product. They find product-market fit with services, then slow everything down by pivoting to build a product and sell seats. Take this as a direct counter-argument: you don't need an externally-sold, per-seat product to build a successful AI business. The market is telling you that the integration layer, the orchestration, the delivery quality are just as valuable as the IP itself.

We previously wrote about the blurring lines between products and services in the AI world. The Fin/Sierra juxtaposition is another illustration of the disappearing divide. Both companies offer customer service agents that resolve problems autonomously without human agents getting involved, albeit one works primarily with SMB and corporate customers while the other goes after enterprise. Both use outcome-based pricing, charging for successful resolution of queries. Yet in terms of margins and sales team composition Fin looks a lot like a traditional software business, and Sierra looks like a high-end consulting firm.

For AI services operators, this means your competitive landscape has expanded. Assuming other business areas evolve the way customer support has, your next closed-lost deal may not be lost to another services firm, it'll be lost to an entrenched software vendor that now has an offering competing with you directly. They'll price it below cost to protect their SaaS seats.

We're already seeing this in AI training. An existing LMS vendor ships a half-baked "AI literacy course." They're already an approved vendor: no lengthy procurement form, no new supplier onboarding. On the surface it ticks every box for "AI training." It's an easy yes for the HR or L&D leader. You never even got the meeting.

Two things you can do. First, go deeper in discovery. Don't assume you're pitching against other services firms. Ask which software tools are currently in use and find out which vendors have started encroaching on your territory. Know the competitive landscape before you walk into the room. Second, if you can't beat them, consider joining them: open partnership conversations with the vendors most likely to compete with you. There may be a gap in their offering that you're well placed to fill, and a partner relationship is a better outcome than a lost deal.