Salesforce's $3.6 Billion Fin Deal Is a Build vs. Buy Lesson Worth Learning

Salesforce already had an AI agent platform. Agentforce reached $1.2 billion in annual recurring revenue this year, growing 205 percent year over year. It still paid roughly $3.6 billion to acquire Fin, the AI customer service company formerly known as Intercom, in the company's fifth acquisition of the year.

The reason points at something worth understanding regardless of company size: Agentforce and Fin solve the same general problem for different customers. Agentforce is built for large enterprises that want deep customization. Fin is a pre-trained, fast-to-deploy agent aimed at companies that need something working quickly rather than something built exactly to spec. Salesforce didn't need Fin's technology so much as it needed the packaged, quick-deploy version of what it already had, and building that internally would have meant competing with its own product roadmap for engineering time.

When buying beats building, even for a company that could build it

Salesforce has the engineering resources to build a fast-deploy AI agent product. It chose not to, and the decision clarifies something useful about when acquisition makes more sense than internal development, even for a company with real technical capacity.

Fin brought three things Salesforce couldn't easily replicate on its own timeline: over 30,000 existing customers already using the product, a resolution rate Fin claims outperforms Salesforce's own Agentforce Help Agent on a comparable metric, and an engineering team that had already spent years optimizing specifically for fast, simple deployment rather than deep customization. Building toward that same outcome internally would have meant years of iteration against a moving competitive target. Buying it meant Salesforce could offer both products, each aimed at a different customer, starting almost immediately after the deal closes.

The complication worth noting

The same month it announced the acquisition, Salesforce cut jobs across its Agentforce, MuleSoft, and Marketing Cloud teams, according to a California WARN notice, though the company said the reductions didn't touch its core Agentforce team. A $3.6 billion acquisition and internal layoffs in the same AI product family isn't a contradiction so much as a real tension every company scaling AI products is navigating: where do you invest in acquired capability, and where do you consolidate the capability you already built.

What this means for a business much smaller than Salesforce

Few companies face a build-vs-buy decision at Salesforce's scale. But the underlying question applies at any size: is the capability you need something your team could build faster and cheaper than buying it, or does the market already have a solution built by people who've spent years solving exactly that problem, at a cost that's actually lower than the engineering time it would take to replicate?

That's a harder question to answer honestly than it sounds, because most teams have a bias toward building, especially engineering-led ones. The businesses that get this right tend to be explicit about the comparison: what does building actually cost in time and opportunity cost, not just in engineering hours, and what capability, credibility, or customer base does an acquisition or off-the-shelf tool bring that internal development genuinely can't match on the same timeline.