If I were given one cent every time I hear the word “AI” at Zoholics 2026, I’d be leaving Houston as a billionaire. Alas, it wasn’t the case, but AI most certainly was a leitmotif of this summit. We were in Houston on May 11th for Zoho Inspire, and on the 12th and 13th for Zoholics USA 2026 — the same George R. Brown Convention Center, familiar hotels, quiet and clean Houston downtown, and a slightly different tone than the one we remember from prior years. In 2025, the question in the hallways was “Are you using AI yet?” This year, the question had quietly become “Did your AI actually work?” — a more interesting question, to be honest, and one Zoho appeared determined to answer in front of us, the Partners, and its own customers, on stage, with the mics on.
What follows is our read of the event as a Zoho Partner. We will not turn it into a feature catalog; the Zoho marketing team is perfectly capable of doing that on its own. Instead, we will focus on what the conference signaled about where Zoho is going as a brand and what that means for the people who actually run their businesses on the Zoho platform.
Zoho AI is All About the Context, Not Models
I think the most important sentence of the entire Zoholics 2026 was delivered early on Day 1, at the Enterprise AI Session Keynote by Ricky Thakrar and Vijay Sundaram, Zoho’s Chief Strategy Officer. On a near-empty slide it read:
AI isn't underperforming because the models aren't powerful enough. It's underperforming because the business context those models rely on is fragmented, ungoverned, and invisible to the systems meant to act on it.
We all know that modern LLMs have inherent limitations stemming from their nature, and there’s no AGI on the horizon. However, as a technology, AI is slowly finding its place in business processes, and I think Zoho’s statement is remarkably fair to say at its own user conference. Zoho is openly conceding, and asking its customers to concede with them, that the bottleneck is no longer the AI. It is everything else [it always was — DS]. So yeas, AI is not a magic wand, but it’s a very good shovel.
Vijay Sundaram framed the institutional knowledge of a business across three places where it actually lives: in your systems (CRM, ERP, databases), in your documents (policies, wikis, email, spreadsheets), and in people’s heads (judgment, memory, experience, and what he politely called dogma). His point was that most of this is invisible to the AI a business has licensed. The model did exactly what it was asked. What it was missing was context.
This framing is not unique to Zoho — it is in the air of the entire enterprise software industry in 2026 — but Zoho is the first platform vendor I have seen put it on a keynote slide. The right question, Sundaram argued, is no longer “do you have AI?” It is “does your platform make context legible?”
That single reframing was what the rest of the conference, in different ways, was trying to answer. I should note that, unsurprisingly to many attendees, most of the questions to the speakers at every panel were… yes, about the AI.
The Honest Talks on Zoholics 2026
The most useful session of the two days was Tejas Gadhia‘s Uncomfortable Truths on Day 2. It was not a product talk. Gadhia, Zoho’s Head of Applied AI, was telling several hundred customers things that vendors rarely say out loud:
- “Zoho doesn’t solve all your AI problems. No single vendor does.”
- “There are no magic agents. There are workflows with different amounts of AI judgment inside them.”
- “The real question is not ‘is this an agent?’ The real question is: what decisions is the model allowed to make, and what happens when it is wrong?”
- “Companies are torching budgets on AI.”
- “Activity is not outcomes.”
I second all of them, especially the second one. Then Gadhia put up a screenshot, tracking how many times the word “agents” would be said in the morning keynotes; the final count was 192 (sic!) The joke had a point: the word has become so promiscuous it has lost most of its meaning. He pulled it back to something useful:
An agent is a workflow with a defined level of judgment, a defined set of decisions, a defined permission scope, and a defined plan for what happens when it is wrong. That is it. Everything else is marketing.
Zach Zivnuska‘s “What It Actually Takes to Implement AI” was the most operational of the AI talks. I’d like to mention these key points:
- The 5x rule. Whatever it cost to build the first version, it will cost five times that to keep it useful after launch. Models decay; organizations change; data drifts. The maintenance is the work. [DS: True. We know that firsthand due to our work on AppVisor.AI]
- Minimum viable data. Field prediction needs at least 75 identical records per picklist value. Lead and deal scoring needs 200 distinct records with a clean split between ideal and non-ideal outcomes. Below those thresholds, models confidently lie. [DS: Also true. Zia in Zoho CRM struggles if the data is not sufficient and becomes literally useless]
Most Zoho customers are already paying for a meaningful amount of “Zoho AI“ (predictive scoring, anomaly detection, sentiment analysis, smart routing, natural-language search). Before licensing anything new, audit what you already own.
The Big Product Announcement: Zoho ERP
The single most consequential product announcement for the US market was Ram Vaidyanathan’s first look at Zoho ERP — “AI-native ERP for the modern enterprise.” This is Zoho’s formal entry into the mid-market and enterprise ERP segment that has historically belonged to NetSuite, Microsoft Dynamics 365, SAP Business One, and a handful of industry-specific players.
Zoho ERP unifies four mission-critical pillars into one platform:
- Core Financials — accounting, reporting, compliance
- HR & Payroll — people operations and payroll compliance
- Supply Chain — procurement, inventory, fulfillment
- Billing Management — subscriptions, billing events, revenue recognition
Day-one industry readiness was demonstrated for manufacturing (finally, with BOMs, work centers, shop-floor visibility, WIP accounting), retail, and distribution, with non-profit support also flagged. The demo included a full manufacturing order workflow for a fictional solar inverter, with job cards and operations breakdown — which is to say, Zoho is not pretending this is a re-skinned Zoho Books. It is a different product, with deeper bones.
What we will be watching as a Zoho Partner:
- The migration path from Zoho Books/Zoho Inventory for existing customers who outgrow the Zoho Finance apps. Zoho was deliberately vague about this at the conference, and that is fair — the answer matters more than the date.
- The pricing position relative to NetSuite. Zoho’s traditional aggressive cost advantage matters most precisely where NetSuite is most painful.
- The implementation discipline. ERP is not a Zoho CRM project. The discovery and process-mapping work is heavier, the stakes higher, and the cost of getting it wrong is measured in months of operational drag. This is not a DIY product.
We will publish a dedicated piece on Zoho ERP as more details become public.
The Other Zoho Product News Worth Knowing
On Zoholics 2026, the Finance & HR track had a productive two days. The product announcements with the highest practical impact for our customers:
- Zoho Procurement — a new source-to-pay platform, presented as the missing piece between Zoho Inventory and Zoho Books. If you have ever stitched a procurement workflow together using Forms, Creator, and approvals in CRM, you know why this matters.
- Zoho travel and cards management, together with an AI-native rewrite of Zoho Expense — a coordinated push into the corporate spending category, with native cards now part of the equation. Watch this space; the unification of corporate cards with expense reporting is where SAP Concur built a decade-long moat.
- Zoho Payments is being positioned as the native payment rail across Books, Billing, Invoice, Inventory, and Commerce — solving the “my cash position is always one sync behind” problem that anyone running an integrated payment processor knows well.
- Zoho Practice received a serious update on the accountant-facing side, with deeper workflows for firms managing many client books at once.
Marketing got a long-overdue Zoho Backstage refresh (bulk exhibitor import, DXF import for venue floor plans), and Bigin continued its march of small, well-judged additions — WhatsApp automation, scheduled email drafts, and the built-in card scanner powered by Apple Vision are all live or imminent.
Zoho AI, Zia, and Zia Agents: What Is Real
Zia is no longer a single AI assistant. It is now a layer threaded through the entire Zoho stack — search, BI, CRM, Desk, Books, Mail, Calendar — with Zia Agents as the named framework for autonomous workflows. The usage statistics Zoho shared on Day 2 are worth recording for posterity:
- 9,000 customers signed up for Zia Agents
- 19,000 agents created
- 13,000 agents deployed
The numbers are modest by Zoho’s overall customer base, and that is precisely the point. Most customers have not yet deployed an agent: because they do not need one, or because they do not know how to do it. The audience for Zia Agents is, at the moment, the early-adopter slice of the install base. For the rest of the customers, the message is less “deploy an agent” and more “notice that Zia is now embedded inside the apps you already use.”
Ramki Rajapandiyan, Head of Data Science for Zia Agents, was careful to frame the agents as digital employees with their own identities, audit trails, and scoped permissions, language we have not historically heard from Zoho, and language we welcome.
The reference architecture demoed in Austin Adair and Naval Patel’s “Extending Zoho with AI” panel, a supervisor model routing requests to domain-specific agents (CRM, Sprints, Desk, Books, Knowledge Base) over Zoho MCP — is a useful mental picture for anyone building or evaluating Zoho AI tooling.
The Zoho One Story
Tyler Bussani’s Zoho One Keynote on Day 2 was, in some ways, the spiritual companion to the AI Session Keynote on Day 1. Where the AI keynote argued that context is the bottleneck, the Zoho One keynote argued that unification — not integration — is how you give a platform that context. The phrasing he used was clean:
Integration is many things connected by lines; unification is many things sharing a data layer, an identity layer, a search layer, an AI layer, and a calendar layer, so that every app already knows what every other app knows.
Anyone who has lived through a multi-vendor SaaS integration project (most of our case studies are exactly this) will recognize the difference instantly. It is the difference between “my apps talk to each other” and “my apps already know.”
This is also where Zoho’s positioning against the “average company runs 101 SaaS apps” statistic (from the cited Okta study) lands hardest. The pitch is not that you should consolidate from 101 apps to one app. The pitch is that the dozens of pieces of your business that live across departments — sales, billing, support, finance, HR, marketing — should share a context layer rather than a list of integrations. That argument was, frankly, more persuasive in 2026 than it was in 2025.
A Zoho Partner's Take
Watching Zoholics 2026 from a Partner’s chair was instructive. A few observations we want to underline:
The unglamorous middle layer is now the differentiator. Anyone can buy a license. Anyone can click enable on a Zia feature. What separates a Zoho deployment that pays its return on investment from one that quietly gets abandoned in eighteen months is the work that does not show up in marketing videos — data hygiene, ownership clarity, governance, role design, change management, kill criteria, and the systematic capture of the institutional memory that would otherwise walk out the door with one departing employee. This is the work Partners do. It is also the work the most candid talks at Zoholics 2026 were openly demanding that customers do.
Audits before implementations. Penumatsa’s “audit before you buy” point is one we already make in our Zoho Audit engagements, long before Zoholics 2026. It is now, finally, vendor-endorsed at the keynote level. If you have a Zoho account that has been live for more than two years and you have not done a structured audit of what AI you are already paying for, what data thresholds you are meeting, and what processes are ready for automation versus the ones that need cleanup first — you are leaving value on the table. That is no longer a Partner’s marketing effort; it is on the Zoho slides.
Zoho ERP is going to redraw the mid-market map (at least we hope so). Watch the next twelve to eighteen months. The right reading of the announcement is not “NetSuite alternative” — it is “the ERP layer Zoho One was missing has finally been built.” For our manufacturing, distribution, and retail clients, this is a meaningful new conversation.
AI is not a project; it is a discipline. Every single one of the honest talks made this point in a different vocabulary. Crawl, walk, run. Ugly pilots. 5x rule. Kill criteria. Augment, do not replace. Measure feel, not accuracy. These are not slogans, they are how grown-up software gets built and kept alive inside a real business.
Where That Leaves You
If you are a Zoho customer who attended Zoholics 2026, you already heard most of this directly. If you did not attend, the short version is: Zoho’s AI story has matured from feature-list to platform-thesis, and the price of admission to that story is the unglamorous work of getting your business context legible to the systems you have already paid for.
The good news is that the work is finite, well-understood, and increasingly possible to do with the help of people who do it for a living.
We are an Advanced Zoho Partner. This work is what we do. If you are thinking about an AI initiative, a Zoho ERP migration (after Zoho ERP will be released, of course), a Zia Agents deployment, or simply an honest audit of where you currently stand — book a Clarity Call and we will tell you what we see.
Until Zoholics 2027!
