What should customers expect from SaaS vendors in an AI world?

Last month was ZohoDay26, our annual analyst event. One of the topics we addressed head-on was what's on everyone's mind — the elephant in the room — whether AI marks the death of SaaS. Recently, diginomica reported on the event and even profiled one of our customers. In this article, I want to give readers a look into how we think software vendors need to alter their value proposition.

Inside Zoho, we are approaching the supposed 'SaaS-pocalypse' with internal rigor and reflection. What follows is an article inspired by a letter that I originally posted on Connect, our company intranet.

Why AI puts pressure on SaaS vendors?

While we readily acknowledge that AI is a threat to the profitability and viability of SaaS providers, we need to clearly understand (and communicate to those that don't) the economics of a SaaS business and what it actually does for customers. This leads us to better predict what a DIY approach by customers to software development can achieve, and what it cannot. We owe it to customers to objectively and transparently lay out the factors they must bear in mind.

Let me start by recapping the simultaneous threats posed by AI today to a SaaS business.

1. Puts pricing (hence margin) pressure on all competitors.
2. May result in less demand for seats and licenses, from customers.
3. DIY trend will make some customers choose build over buy.
4. Massively lowers the barriers to entry for new competitors, who can now build rapidly.
5. Business model pressure by forcing trends like outcome or results-based pricing.

It will also create huge opportunities for vendors who can adapt this technology to create even more value for customers through aggressive pricing, handling the risks that come with AI and by providing the right guardrails for AI-built apps.

So, what should customers expect from software vendors?

I will bring this down to four points that true software partners must understand and articulate well.

1) Strengthen credentials as a value player

Delivering value is paramount. In our case, Zoho has always been known as a value player, starting with price, but not confined to this. But that means we have to be good at two things here. First, we need to understand what value really means to the customer and communicate this well. Second, we need to maniacally focus on driving customer value in everything that we do. This means engaging with customers at all levels, from product development through sales and marketing to customer success.

While advances in AI threaten all vendors, value players are more likely to withstand the assault. Vendors with high prices, who have been skimming margins off the top of the largest companies, and who have been accustomed to operating under fat margins and non-discriminating customers, will hurt the most. Many of the same customers will now turn to discrimination and look for options. We must be ready, not just with price, but with capabilities and an explanation of our strategy.

We need to know how to demonstrate our value convincingly.

2) Underpin the strength of a system of record

A true applications platform serves as the customer's System of Record. An SoR is a place where reality is defined, validated, and made operational. The last part gets increasingly important under AI. In a 'vibe-coding' world, AI dramatically reduces the cost of software development, automation, and integration. However, it dramatically increases the costs of compliance, trust, institutional memory, and detection of maverick and risky action.

A true SaaS platform should provide the 'guardrail'' for AI action, through appropriate authorization validation. SaaS customers serve their own customers and need to be accountable for their data integrity and trust. Accessing operational AI through a well-governed system of record, rather than coding around it, affords customers better protected, more contextual, and broadly applicable use cases, which can be implemented immediately.

3) Advance customization through a platform approach

Platforms trump point solutions. Over the last few years, Zoho has been building a platform approach that allows customers to build the exact capabilities they need for their specific businesses, by leveraging our stack — all the way up from cloud infrastructure, to building blocks and modules, and to the applications themselves. The idea is that customers get bespoke capabilities by reusing and building on top of tested and verified capabilities that we provide, within applications modules, workflows, builders, and tools.

For customers, this is an unbelievable promise — completely custom software, at the price and reusability of 'off-the-shelf' systems.

Our current approach with AppOS takes this even further, as we will see this year. We will create a system that allows customers to DIY software any way they want, through pro-code, low-code, or vibe-code options. You will hear more on this as the year progresses.

4) Without unified business context, AI and agentic technology can't change outcomes

Agentic approaches are here to stay. They will allow users to define their own user interfaces and workflows, that hitherto was buried in the application layer. If SaaS vendors want to maintain deep relevance, they must create value for businesses by first owning then decoupling data, business context, and AI model layers.

Yes, that means providing a broad set of tools that capture both structured and unstructured data from internal and integrated third-party systems. It also means ensuring the governance, security, privacy, and compliance of this data that's otherwise known as 'designed knowledge'. Designed knowledge is a critical component of delivering context to customers, but so is the "discovered knowledge" (signal data that comes when a client opens or ignores an email, edits or shares a document, etc.) that Zoho's SoR captures. Combining designed knowledge with discovered knowledge with our system of record produces valuable unified business context for customers.

Deploying that business context across a choice of AI models is crucial (In Zoho's case, that means public, hosted open source, and Zia LLM, Zoho's own model). When you build your own LLM, you can tune for differentiation. You can custom train and custom deploy for customer requirements. Combining a focused model trained on private, real-time data with a public model providing universal business context gives customers relevant value from the technology.

Therefore SaaS vendors must embrace the agentic world. That means building a broad set of agents for our customers, across all systems of record. That means empowering customers to easily create their own agents using any tool they choose (eg, Claude Code). They may also bring third-party agents to query your SoR.

The world is changing fast. If SaaS vendors don't adapt, then trouble lies ahead. Customers are the true arbiters of value — earning their trust is the only moat a vendor can really have.