ZohoDay 2026 - SaaS may not be dead, but the customer value question is burning hot. Why Zoho's leadership sees the market differently

AI fatigue has given way to "SaaS is dead because of AI" fatigue. Deep breaths all around - I believe the markets are overestimating agentic AI's ability to build deterministic/auditable software at enterprise scale.

But: the markets aren't wrong about this: SaaS vendors need to do a better job of justifying their value. 

Let's face it: some SaaS vendors deliver on that price/value scale, and many others don't. And yes, agentic AI, properly harnessed, does have a notable impact on software delivery. 

But as we saw clearly at ZohoDay, 2026, established vendors can apply those same AI development tools - perhaps more effectively, given the domain expertise they can bake in, on top of deterministic workflows customers already rely on. 

At this potent juncture, Zoho is one of the best software vendors you could talk to. Why? They are always knee deep in technical experiments, which extend to the entire AI infrastructure stack. And they conduct these experiments amidst a profitable model, without quarterly number pressure/distractions. 

But as we saw once again last week, Zoho can also be counted on for provocative stances ("We aren't laying anyone off"), and new framing ("this is the delayering of the stack"). Zoho's original thinking brings zingers that change the scope of the debate, e.g. this scorcher from Zoho Chief Evangelist Raju Vegesna: 

If someone can pull the plug on intelligence, are we really sovereign?

Vegesna was speaking of the global trend of data sovereignty - and localized models that break dependence on a single LLM/intelligence provider or hyperscaler. 

SaaS was more than a distribution model; it was "an institutionalization of risk transfer."

As the analyst event opened, Zoho launched into an incisive review of SaaS market risk from Vijay Sundaram, Chief Strategy Officer at Zoho Corporation. Sundaram sees five market risks SaaS vendors are now facing: 

• Pricing pressure/margin pressure ("This is something we expected as a company for a long time")
• Less demand (reduction in seats/licenses?)
• DIY AI trend for customers
• "Massively lower barriers to entry for new competition"
• Business model pressure

 Here's the jugular twist: SaaS vendors are feeling the pressure from all at once. Sundaram:

So the interesting thing is that all of these are happening simultaneously in the market. So where does this leave us, and how do we look at things? And how does this impact the future of SaaS companies?

Sundaram's takeaway? This is about a shift in risk: 

SaaS is thought about as a way that you price and distribute software. When you think of SaaS, the first thing you hear is that, 'Oh, pay by the month; you pay as you go. So there's a pricing and distribution strategy... I want you to actually think a little bit more. SaaS has actually been a fundamental way where risk got reallocated to the vendor. 

He explained: 

What is customer risk under a SaaS model? Obviously, there's economic risk. Nobody wanted to capitalize expense. They push it back on the vendor, operating expense and all the other stuff... The customer used to own this risk. Second is security and compliance. There's all kinds of security, privacy, and compliance capabilities that are increasingly needed in the modern enterprise world. All of these things are now back largely to the vendor. That's what happened with SaaS.

So we've got economic risk, security and compliance, operational risk (hosting, scaling, uptime, SLAs), provisioning and upgrades. The only risk not traditionally taken on by the SaaS vendor? Customization: 

When customers [traditionally] bought software, they spent a lot of money customizing. They inherited technical debt, because they now divert from where the vendor was and where the market was from.

Ergo: 

SaaS was more than a distribution model, it was an institutionalization of risk transfer. 

"Are customers prepared to take up this risk?"

So what happens next? No one knows for sure - but there are implications for customers and vendors. Zoho believes their overall pricing/value approach to software - along with their full stack approach to AI -  can surpass now-heightened customer value expectations. As for customers, Sundaram says: 

If you take software into control, the risk is also moving back to the customer. Risk is now being internalized. 

Customers that choose to build AI apps will balance that risk with the perceived opportunity: 

What is implicitly happening under this assumption is risk will get recovered. Many customers will be ready to do it, and many others won't... Are customers prepared to take up this risk? You have to build a lot of these operational capabilities in a post SaaS world. 

But that raises inescapable questions for software vendors: 

What types of vendors are more vulnerable? Horizontal vendors, or vertical vendors? Enterprise? Mid-market focus? Hard to tell, right? I don't know the answer to all of these questions, but these are the ones to ask; these the ones to plan against.

Software value might be going down, giving customers operational flexibility has not

Which leads to the inevitable question: what is Zoho doing about it?  Sundaram says value is now about operational flexibility, not software alone. It's about interoperability, platforms - and giving customers the ability to customize to quickly adapt to market pressures/opportunities: 

When value means it's more than price, it's operational flexibility. It's the fact that we build suites; it's the fact that we develop platforms. It's the fact that customers build more and more capabilities at the same price. It's giving customers the freedom to customize these applications. 

All of that is valid. This notion of compressed margins has been something Sridhar talked about 20 years ago. He's floated the idea inside the company that one day, software will get commoditized. It may be free, and in a sense, we built a company under that model.

Getting AI right is about context - Zoho wants to redefine the system of record

Which leads us AI context. In enterprise circles, "context engineering" has moved from fancy/pretentious-sounding buzzword to AI engineering discipline. But recently, the "context" discussion has shifted, via viral arguments that for proper decision/actions, AI needs a context that goes beyond static data, into real-time streams that contain signals that influence everything from purchasing behavior to buying decisions ("signals" could come from workplace chats, account history logs/integrations, real-time social sentiment, and the list goes on. But the needed context for AI certainly doesn't reside in just one conventional database. I agree with that reframing of context, but I do see some problems - more on that down the line. 

Sounds like Zoho been moving in this direction for quite some time (Zoho announced its Unified Context Layer at ZohoDay 2026). Sundaram argued for the relevance of systems of record, while redefining that term:   

 And so here we are today. We are best prepared for this because we built our company around the notion of delivering value. The corporate system of record - I talked about magnificence in the mundane. So we built these capabilities. This is not just the data. It's all the innate knowledge of a customer that's embedded in the system - data, the workflows, conversations, the unstructured data, all of the things that's built up, the context that AI will need to work through as we get into that world. 

So we will be that system of record that is beyond just the data, that captures all the context that is the foundation, the permissions, all of the capabilities that are needed for AI to function above.

For Zoho, this is now about a platform play that puts partners on the same level as internal developers. Sundaram: 

 You'll hear the next day or so about our platform approach, a lot about what we're doing. It's advanced a lot over the last couple of years, and it's been in the works for a number of year before we were put n a position to announce it. The idea is to open up an ecosystem that can build all kinds of sophisticated applications on the Zoho platform in exactly the same way as the next Zoho team would be building an application on top of Zoho. To make that kind of capability available to the ecosystem, so it proliferates these applications that we expect -  built by third parties.

To pull this off, Zoho will need to bring together data, automations, and AI agents that extend beyond Zoho: 

The last part is to enable that automation.Customera are going to use agents. We're going to put them all over our platform. We're going to embrace agents of others, and allow customers to have that type of look and feel they want.

My take - taking up the customer challenge with Shridhar Vembu 

Vegesna's quote about intelligence-being-unplugged sticks with me. I believe open source models will ultimately drive (most) LLMs towards commodity status. This should free up chances to experiment, and reduce intelligence dependencies.

Zoho itself argues that smaller models with specialized context are superior to giant frontier models, at least when it comes to enterprise workflows (to that end, Zoho has now built its own LLMs, in a range of sizes, and 'rightsizing' of models for purpose is already a Zoho specialty). 

Frontier models might grab the sexy benchmark headlines, but as Sundaram says, informing models with the right data mix is indeed the "magnificence of the mundane." Here, data sovereignty absolutely matters. It's the intersection of context and sovereignty that makes the data conversation so interesting right now - and Zoho is smack in the middle of it. 

Zoho has its share of challenges. From Zoho's ERP ambitions to its pursuit of a full AI stack, there is a big scope in play here. Zoho makes the argument that for SaaS to deliver value, flexibility is crucial (agreed), but that customization is key to that agility. But can you enable customers to customize at scale without eventually being hindered by technical debt? That's a potent question I want to hear more from Zoho about. 

I'm more bullish on industry partners extending core apps on an apps/data platform, with imaginative new industry apps, and AI-embedded app ideas, with built-in security/governance provided by the vendor's app framework. Zoho thinks so too, but how effectively will that play out with partners - and with third party integrations? Those topics got some attention at ZohoDay, but they are questions to track. 

One challenge Zoho isn't phased by? AI code experiments. Since his move to Chief Scientist, Shridhar Vembu and team have been pushing on AI coding possibilities. During a virtual keynote at ZohoDay, Vembu confirmed that significant advancements have been made in Zoho's speed/efficiency of software delivery; he cited an example of a compiler generated by two engineers using AI prompts in two days - one that would have taken these two engineers three months without AI tools (right now inside Zoho, Claude Opus 4.5 and Opus 4.6 are the best performing AI coding models, but Vembu largely expects future models to have "different personalities," but "similar capabilities." I'll share more on how these two engineers did this in an upcoming piece). 

But with Zoho pressing this pace of change, how do customers keep up? I asked Vembu: what would he say to his own customers right now? 

Vembu's overriding point? The cost of software must come down: 

I would say today, a lot of the software has to be much more affordable. 

Customers should expect more, says Vembu - and yes, that includes Zoho's.

Expect more value from your vendors, including Zoho. That's what first I would say. Second, in their own business, I would love them to have at least one person experimenting with these tools and see how it works. Be open minded. Don't assume that it's a great help for their specific domain. The [model] may not have had sufficient training for it to help them, but they must find that out experimentally. 

Have somebody who is doing this exploration, that's what I would advise, like how we have been exploring all the time. I see a lot of people all the time using things, giving me feedback, and I jumped on it once I saw some of it myself. So I think that is what I would advise customers to do as well.

But what about the supposed disruption to skills and labor costs? How is Zoho grappling with that? Vembu's next answer didn't hold back: 

I'll tell you one thing I've told our employees: we will absolutely not resort to layoffs. Others are like, 'Our ship is burning; everything is going down,' which is not happening. We are doing well. So I said we will not do proactive layoffs to prepare, all of that nonsense. We will not do it... I wanted to take that off the table. 

I said, 'We will adapt. We'll re-invent, but we are not going to cut - we are going to offer the customers a lot more.' So we are hiring very slowly. We are not hiring very many software engineers right now, but we are going to keep what we have, because I don't want to transfer all the uncertainty to the employees. The business is still doing well, so I don't want to sound alarmist on all of this - but I'm just highlighting that the ground is shifting. 

Mani Vembu, CEO of Zoho, captured the challenge ahead: 

B2B platforms are complex; their business models are not sustainable... And so the future really looks unclear. We want to look at ourselves and say, 'Is there a better way to build, run and evolve business software?'

That's the burning Zoho preoccupation, and it's the right one. I have a slew of content to flesh this out, including videos, AI deep dives, and a ZohoDay show wrap podcast with Brian Sommer, embedded below. Let's see where this goes.