Ask a sales manager which deals closing this month have gone cold, and the answer isn’t in one place. The pipeline is in the CRM, the open tickets are in the support desk, and the overdue invoices are in the finance system. Getting the answer means three logins, a few filters and a spreadsheet, and by then the moment to act may have passed.
At its Zoholics user conference in Chennai, Zoho launched Zia Chat, a conversational AI layer that lets people ask that question once, in plain language, and get a single answer drawn from across the business. It arrived alongside seven industry-specific applications. But the chatbot is the last step of Zoho’s strategy, not the first.
The first step, according to Anand Nergunam, Global Vice-President of Revenue Growth at Zoho Corporation, is an admission. Zoho does not necessarily have the best knowledge of how a particular industry works. Its answer is to learn it from customers, build with them, and turn what it learns into products that the next customer can buy.
Nergunam is candid about what Zoho does and does not know. “What we are good at is understanding code, how to write code, how to create user interfaces and how to productise something,” he says. “What we do not necessarily know is how an industry operates.”
Take an NBFC customer. Zoho may not initially know “how loan processing happens, what checks and approvals are required, what the RBI requires, how underwriting works for a particular loan or insurance product, or how deduplication is handled.” So the company, he says, approaches customers “as humble students”: “You are the customer, so tell us what you need. We will create it together with you.”
The customer gets to be first to try the solution, tailored to their world. Zoho comes away with something it can reuse. In automotive, “we started with Mercedes-Benz, then worked with Force Motors, and today we work with several other customers.” The process, he says, is simple to state: “put your heads down, learn, productise and then go to market.”
That learning includes unglamorous operating realities. Many people in NBFCs work in Tier 3 or Tier 4 locations where connectivity is limited, so applications “may need to work on 2G speeds or offline.” “We could sit in an ivory tower and build something, but then customers would be forced to adapt to us,” he says. “We want to flip that model and build around their needs.”
Why the old models fell short
Nergunam places the strategy in a longer history. In the 1980s, businesses relied on home-grown software. “These solutions were highly tailored and integrated, but they were difficult to change,” he says, and they were often not equipped to deal with evolving security vulnerabilities.
Then came the era of “best of breed”, where businesses picked the best CRM, the best accounting software and the best tool for each function. “The problem was that these applications did not necessarily talk to each other, creating fragmented systems,” he says. “They could be customised to some extent, but they often did not speak the language of the industry or relate to the business in a meaningful way.”
The result is a market in which “companies buy different applications and assemble them superficially.” His case for Zoho’s approach follows from that: “When software is truly tailored to a business, efficiencies can potentially improve significantly.”
What co-creation delivers to the customer
Nergunam says the benefit is a balance between the control a business wants and the efficiency it needs. Automotive shows the tension. An OEM wants the application experience to be identical across its dealer network, because “the OEM is the brand, not the individual dealer.” Yet dealers are wary of sharing lead data with the core system. A dealer who collects leads at a roadshow may not want another dealer to see them.
“So, you need control over the application while maintaining freedom over data ownership,” he says. “Every time an update is made, all dealers receive the same update, while data ownership remains with the individual dealer.”
In financial services, the pressure is speed. “The market is extremely dynamic, competition is changing and customer expectations are evolving,” Nergunam says. He describes a customer who can “take a picture, have the system identify them, retrieve the relevant information and provide an in-principle approval within 30 minutes.” If a lender has to rely on an external vendor for every such change, “it could take six months. By then, the market may have moved on.”
How this differs from a custom application project
The obvious objection is that co-creation sounds like a bespoke build, with all the cost and lock-in that implies. Nergunam draws a line. “In a traditional custom application project, the customer owns everything,” he says. “In our model, the customer owns the workflow and the UI, while we take care of the scalability, components, complexity, engineering and integration.”
Pre-built modules and integrations come out of the box, and customers extend them on Zoho’s own low-code platform. A workflow can be built in a drag-and-drop editor. An approval mechanism that once needed someone to modify code becomes an editable visual workflow. The aim is to make tooling simple enough “to create a centre of excellence within the customer’s organisation.”
The payoff, as he sees it, is that a business owner can adapt software without going through “an IT team, procurement team and RFP process” and waiting for an external vendor. A product launch that previously took two months “could potentially be reduced to around a day and a half.”
Zoho’s MCP support extends this to natural language. A business user can say, “I want to create a new module” or “I want to edit this workflow,” and “the system understands the underlying architecture of the CRM.”
That flexibility is also what makes CIOs nervous. Customers, Nergunam says, are “increasingly concerned about shadow AI and shadow IT.” Leaders “are looking for vendors that give them the flexibility to adapt software to their own world without introducing the risks associated with that flexibility.” His shorthand for the goal: “bespoke software without bespoke risk.”
Then comes the AI layer
Co-created industry software is the foundation for Zia Chat. Zoho describes AI maturing in three stages: generic AI that understands language, application-aware AI that understands individual apps, and industry-aware AI that understands entities, rules, workflows and relationships across the business. Its position is that AI acts effectively only when it understands the industry, the workflow and the business context.
Zia Chat is built natively into Zoho’s applications, where identity, permissions and business relationships already exist. Zoho says it uses semantic search, a “work graph”, resolvers and relationship traversal to work out how customers, deals, tickets and transactions connect. MCP integrations let organisations bring in information from systems outside Zoho.
Nergunam links the two layers. Take the automotive dealer management system, which is “deeply interconnected with the customer’s own tools as well as Zoho’s tools.” “On top of that, we can put something like Zia Chat, which understands the data and its sources while respecting the guardrails that have been put in place,” he says. “This can help businesses become truly future-ready.”
The guardrails are built into the product. Zia Chat can only access the information and actions available to the user: if you can view a record but not edit it, neither can Zia Chat. Users review and confirm actions before anything is created, changed or sent, answers can link back to source records, and data accessed by Zia Chat is not used to train Zoho’s models. On cost, Zoho co-founder Sridhar Vembu has said the aim is to cut the tokens AI workloads need, use smaller or local models where possible, support customers’ own model providers, and bundle the AI into its software.
Depth over breadth, and a legacy opening
The seven new applications cover retail (RetailIQ), automotive (AutoDMS), banking and financial services (Zoho LOS), healthcare (MedScribe), education (Zoho Campus), real estate and restaurants. Nergunam says the ambition is deliberately narrow. “We want to own industries,” he says. “Our goal is not to touch 70 industries superficially. I would rather go deep in seven industries. The depth is where we create value.” He calls these “micro markets, but they are not micro in terms of their size.”
He expects the co-creation model to compound. Once Zoho has strong implementations in a vertical, “we start developing significant authority in that industry,” and word spreads. “For us, creating two or three strong success stories can create momentum.” Asked for customer numbers in automotive, he said the team would share them, noting that with OEMs it “is not going to be in the hundreds.”
The larger prize, he argues, is the installed base of broken software and paper processes. He recalls a hotel whose screens looked like they were from the 1980s. “These systems are broken and dated, and they do not talk to each other,” he says. “What will an agentic AI system do when you are working with tools from the 1980s?” Workshops still run on pen and paper, and in hospitals he was “shocked by the amount of information nurses still write by hand.” For Zoho, he says, this is “a goldmine,” because it already has horizontal applications and can put a vertical layer on top.
He is careful not to pitch it as a cost play. Many businesses, he says, “are not ready to use AI at all because of the legacy systems they rely on,” which may lack the right APIs or guardrails. The case is readiness: “The value that this can generate for businesses can far outweigh the cost they incur by paying for the software.”
Asked what Zoho wants to be in a year, Nergunam says: “the vendor that helps businesses harness the true power of AI.” That, he adds, “can happen only when businesses have interconnected systems that talk to each other, build deep context and minimise risk for customers, while delivering a cost and value proposition that makes sense to them.”
The bet underneath it is that AI is only as good as the context beneath it, and that context is learned by sitting with customers first.