AI hits stride at ZohoDay 2026: AppOS, agents, and low-code with guardrails
Zoho marked its 30th year with a clear message: scale with control. The company has crossed a million customers and is no longer the challenger brand. It's holding ground against SaaS titans while pushing back on unchecked AI agents that threaten to fracture enterprise stacks.
Last year brought Zia's cross-app natural language interfaces and an agentic builder. This year focused on integration, reliability and operating discipline - so humans and agents can ship value without creating chaos.
AppOS: a reliability layer for agentic work
AppOS extends Zoho's "operating system for business" approach across 50+ apps and thousands of integrations. The premise: most companies already run on hundreds of apps, and "adding agents on top" without a common layer increases breakage and risk.
"Most of the customers I meet say they already have more than 200 applications, and they have a hard time changing functionality fast enough to meet business demands, so they end up only adding to the complexity," said Chief Executive Mani Vembu. With MCP servers and coding agents (think Replit and Lovable) in the mix, "the foundations of our apps develop cracks and become very weak. We need a stronger reliability layer to support differentiation."
AppOS aims to unify semantic integration, development and data, so people and agents can build custom apps and connect services and models - with governance, privacy and sovereignty built in.
Will AI agents kill SaaS? Not likely - but consolidation is here
Some predict a mass extinction of SaaS. A more realistic view: a sharp consolidation cycle. A recent survey cited 264 SaaS apps falling into shadow IT, with only 26% of purchases managed by IT. Deloitte expects 40% of agentic AI projects to fail by 2027. The lesson: outcomes and integration win; hype does not.
"AI has moved from experimentation to operationalization within a year, so we're starting to see best-of-breed consolidation," said Vijay Sundaram, Zoho's chief strategy officer. "Investors, stockholders and business stakeholders need to tie AI projects to outcomes through integration. Your systems of record, APIs and all of the layers of your technology have become strategic assets, and now there is some magnificence in what was mundane."
Architecture first: let agents work inside governed systems
Core systems such as CRM and ERP, plus low-code builders like Zoho Creator and Zoho Flow, become tools agents can use under business rules. Developers don't disappear; they shift from grunt work to domain-led design and customer experience.
"What we're doing with our own engineers is less coding, more domain expertise and more entrepreneurial, creative thinking about what features would benefit customers and the business," said Mo Umerji, head of enterprise architecture at Newcross Healthcare Solutions. "With off-the-shelf creators, you don't have the ability for one custom application to talk to another. Can we really trust the tools to keep the data where we need it? Especially for healthcare, you can't have an agent or an LLM training other people on your data. With Zia AI and Zoho, our data stays in a closed ecosystem."
Right-sized AI: SLMs, constraints and pragmatic infrastructure
Run the smallest model that gets the job done. Zoho is pushing small language models specialized by domain, plus system-level constraints to keep costs and latency in check. "We're introducing system constraints into our AI logic, so our customers can rightsize models within the context of their work for better cost and sustainability," said Ramprakash Ramamoorthy, director of AI research at Zoho Labs.
Under the hood, Zoho continues to build out global AI data centers, a Postgres-style distributed data lake and "flex" inference that can run Zia or third-party models on commodity GPUs for low-latency performance. The company also previewed a maxed-out in-house server prototype for highly regulated environments.
Partners, verticals and sovereignty
Channel revenue is growing quickly as global SIs and local MSPs deliver domain-specific solutions across Asia, Africa and Latin America. Vertical Studio lets partners build tailored offerings - from legal firm operations and maritime management to AutoDMS for dealer workflows and a white-labeled platform for an international janitorial franchise network.
"Zoho allows us to not only change our system, but to scale to our customer's specifications within Vertical Studio," said Catherine Carignan-Turcotte, chief business development officer at NSI Solution. "We can configure their apps with reduced functionality for just what franchises need to use, and push updates with fast deployment."
Zoho's privacy-first stance aligns with a growing push for data sovereignty. "If somebody else can grab my data, or pull the plug on my messaging, or hijack my payment system, am I really sovereign?" said Sundaram. Tighter integration with the ManageEngine ITOM/ITSM suite also enables white-labeled operations and security for MSPs supporting larger enterprises.
What executives should do next
- Treat integration as strategy: Inventory systems of record, APIs and data contracts. Fund integration work equal to feature work.
- Define an agent operating model: Decide where agents assist, where humans approve and how you audit outcomes. Capture prompts, policies and logs.
- Right-size models: Prefer SLMs and task-specific agents. Set cost, latency and accuracy thresholds per workflow.
- Reduce shadow IT: Consolidate duplicative tools. Use AppOS-style guardrails to connect what stays.
- Build with domain partners: Use vertical studios and MSPs that can own outcomes, not just hours.
- Plan for sovereignty and resilience: Choose vendors with clear data boundaries, regional hosting options and viable on-prem/edge paths if required.
Key questions for your next leadership review
- Which business outcomes will agents improve this quarter, and how will we measure them?
- What is our reliability layer for prompts, integrations and data lineage across apps and models?
- Where can SLMs replace heavyweight LLM calls without losing quality?
- Which partner can deliver a vertical solution faster than we can assemble internally?
- How are we enforcing privacy, sovereignty and model-training restrictions on our data?
Zoho is leaning into context as the differentiator - connecting apps, agents and data so teams can ship value faster with fewer surprises. As one executive put it: "Zoho is our force multiplier," said Jeff Anderson, vice president of IT at INTEGRIS Credit Union. "With 80% of our customers never coming into the branch, we are trying to understand our customers where they are. It's not best-of-breed that we're looking for anymore, it's applications that can talk to each other and integrate new processes that can add customer value."
For more strategy guidance on building AI operating discipline, see AI for Executives & Strategy.
Your membership also unlocks:
