Adviser: bolting on AI can cut what a buyer will pay
Itay Sagie argues that copilots, model integrations and third-party AI tools often read to an acquirer as vendor dependency rather than innovation. His test is whether the AI leaves behind an asset nobody else can copy.
Boards keep treating artificial intelligence as a valuation enhancer. Itay Sagie, a strategic adviser to technology companies and investors, argues in a guest commentary for Crunchbase News that in some cases it does the opposite.
His case is that AI can reduce differentiation, compress margins, complicate due diligence and make a company harder to buy. He sets out three places where the damage shows up.
The first is architecture. Startups have been adding AI copilots, model integrations, orchestration layers, prompt libraries, vector databases and third-party tools across the business, which helps teams ship faster.
On Sagie's account, a buyer looks at the same stack and asks a different set of questions. Which models are embedded in the product, which vendors are critical to delivery, where customer data flows, how outputs are monitored, and what happens if pricing changes, APIs break or regulation shifts.
What the founder calls innovation, he writes, the buyer may read as integration complexity, vendor dependency, compliance exposure and security risk. That matters most to strategic acquirers who have to fold the target into a bigger platform.
The second is proprietary data. Sagie's point is that a year ago an AI feature could create excitement on its own, and now summarisation, search, chat interfaces, recommendations, content generation and workflow assistance all run on the same underlying models and infrastructure.
Acquirers rarely pay a premium for integrating the latest model, he argues. They pay for what they cannot build themselves: proprietary datasets, unique customer workflows, distribution, deep vertical adoption, or network effects that improve with scale.
The question he puts to founders is whether their AI work is creating a defensible asset or features a competitor can copy within weeks.
The third is the buyer map. Cybersecurity startups used to sell to cybersecurity vendors and vertical SaaS companies to vertical rivals, and Sagie says AI is moving those boundaries as platforms push into adjacent markets they previously ignored.
His examples: an infrastructure company buying an identity platform because AI agents need secure access controls, an ERP vendor buying workflow automation, a data platform buying a vertical application because domain-specific data is worth more.
His practical instruction is a schedule, not a sentiment. Revisit the buyer map every six to 12 months, because the most logical acquirer today may not be the one that looked obvious a year ago.
For a founder mid-build, the cheap version of this advice is documentation: know which vendors you cannot survive losing, and be able to show where customer data goes. That is a diligence question you will be asked eventually, and it is easier to answer before the stack grows.