The SaaS Model Has a Problem

Salesforce's technical infrastructure is genuinely excellent. The object model, the automation engine, the formula layer, the integration architecture, these are serious tools built over twenty years. Nobody is replacing that with a weekend project.

What is collapsing is the layer above it. The opinionated interpretation of how a business should work, encoded into a data model and sold as best practice. What a lead is. What an opportunity stage means. What belongs on a case versus a contact. Salesforce did not invent these concepts neutrally. They made decisions, baked them into a product, and charged organisations to adopt those decisions as their own. For two decades that was a reasonable trade. Building a custom alternative was expensive, slow, and risky. Paying for someone else's opinion was cheaper than forming your own.

That trade is unwinding. A database structured around how a business actually operates, combined with an AI interface that carries genuine operational knowledge and responds in plain language, now does what the opinionated layer was doing. Better, because it reflects the actual business. Cheaper, because the infrastructure cost is a fraction of enterprise licensing. The AI is not the product bolted on at the end. It is the intelligence layer, trained on internal knowledge, replacing the prescribed interpretation the vendor was selling.

This is not a Salesforce problem specifically. It is a structural challenge for any SaaS business that sold intelligence as its product. The infrastructure underneath is still valuable. The opinion on top of it is not, when the opinion can now be generated fresh, specific, and free.

Agentforce is not the answer. Layering AI onto a data model that was already the wrong shape produces faster navigation of the wrong structure. The innovation required is harder than that, and more threatening to the business model that built the company.

The price of the opinion was always the real product. That product has a competitor now.

Q: What is meant by the opinionated layer in enterprise SaaS?

A: Enterprise SaaS products like Salesforce encode a vendor's interpretation of how a business process should work into the product itself. Standard objects, prescribed workflow structures, and built-in assumptions about relationships and stages reflect decisions the vendor made on behalf of all customers. Organisations historically adopted those decisions because building a custom alternative was prohibitively expensive. That cost barrier has fallen significantly with modern database and AI tooling.

Q: Why is AI a structural threat to SaaS rather than just a new feature?

A: SaaS businesses that sold intelligence, the prescribed interpretation of process, best practice data models, opinionated workflow structures, face a specific challenge when intelligence becomes ambient and cheap. The infrastructure underneath their products retains value. The interpretation layer on top of it does not, when that interpretation can now be generated specifically for a business at negligible cost. Adding AI features to an existing product does not resolve that structural tension.

Q: What does an AI-native alternative to enterprise CRM actually look like?

A: A database structured around how the specific business operates, rather than a vendor's interpretation of it, with an AI interface trained on internal operational knowledge. Users query, update, and interact with data in plain language. The system reflects actual relationships and processes rather than prescribed ones. The infrastructure cost is substantially lower than enterprise licensing, and the intelligence layer improves as it accumulates organisational context.

Q: Is this argument specific to Salesforce or broader across enterprise software?

A: Broader. Salesforce is the largest and most visible example because the gap between its licensing cost and viable alternatives is now significant, but the structural challenge applies across any SaaS category where the vendor's value proposition was primarily the opinionated interpretation of how a process should work. CRM, marketing automation, and service management platforms all face versions of the same question.

Q: Should organisations on Salesforce consider switching now?

A: Not reflexively. Mature implementations carry real value in configured automation, integrations, and accumulated institutional knowledge. The switching cost is genuine. The more pointed question applies to organisations evaluating a new implementation or facing significant licence increases, where the alternative that had no compelling answer three years ago now does.