Humanizing AI: How Zoho Is Building Intelligent Business with Trust, Governance, and Long-term Value

Tejas Gadhia, AI Evangelist, Zoho, in conversation with Krishna Baidya, Senior Industry Director, Frost & Sullivan

Artificial intelligence has become the defining business conversation of our time. Yet while experimentation is widespread, true enterprise-scale transformation remains complex. In this Movers & Shakers dialogue, Krishna Baidya of Frost & Sullivan speaks with Tejas Gadhia, AI Evangelist at Zoho, about cutting through hype, enabling practical AI adoption, safeguarding governance, and preserving the human core of innovation.

“The goal should not be to automate people away. It should be about empowering them to be more productive.”

— Tejas Gadhia, AI Evangelist, Zoho

Moving Beyond the Hype

Krishna Baidya: We are living in a time when everything is about AI. As the AI Evangelist at Zoho, how do you define your role, and how does Zoho approach this landscape differently?

Tejas Gadhia: The role is intentionally a little blurry, and that’s by design. For us, it’s less about chasing hype and more about focusing on practical use cases that genuinely help customers. We avoid flashy demonstrations that are impressive on the surface but lack real business value. Our customers consistently look for high functionality at a reasonable price, and when it comes to AI, they are not interested in gimmicks. They want to know how it will increase productivity, grow their business, generate better insights, and strengthen connections within their teams.

The AI landscape is evolving rapidly, both in terms of models and vendors, and there is enormous pressure to keep up. But Zoho has always taken a patient approach. Over the past 30 years, we have focused on predictability in pricing, security, governance, and long-term value. Our philosophy is simple: build something reasonable, responsible, and grounded in reality. AI must serve business outcomes, not marketing narratives.

Bottom-up Innovation vs. Top-down Mandates

Krishna Baidya: Many organizations are still in pilot phases, struggling to scale AI initiatives. From your experience, where do you see the biggest challenges?

Tejas Gadhia: The biggest challenge is often how AI initiatives are introduced. Many organizations attempt top-down mandates where leadership instructs teams to “use AI,” but that alone is not a strategy. Real value emerges at the ground level, where context lives. Employees understand their daily pain points, repetitive tasks, and inefficiencies far better than executives several layers above them.

When AI adoption is driven from the bottom up, individuals experiment within controlled environments and identify meaningful improvements. This approach tends to be more successful because it aligns with real operational challenges. I also see a distinction between personal productivity and organizational productivity. Individuals are advancing quickly in using AI to make their own work more efficient but scaling that across an organization is slower because it requires governance, shared culture, and alignment.

Leadership’s true role is not to impose AI usage but to remove barriers. They must establish governance structures, enterprise agreements, data policies, and guardrails so employees can experiment safely. Without that framework, people will resort to shadow AI, using personal accounts and potentially exposing company data. Scaling responsibly requires both empowerment and oversight.

Governance, Risk, and Intelligence Sovereignty

Krishna Baidya: Governance and risk often get overshadowed by excitement. What advice would you give organizations to balance innovation with risk mitigation?

Tejas Gadhia: Risk increases significantly when organizations have fragmented technology stacks. Each product has its own privacy settings, governance policies, and security controls. Creating a unified framework across all of them becomes complex and time-consuming. If governance is not prioritized at the top, individuals will find their own workarounds, often through personal accounts, which introduces data leakage and compliance risks.

Organizations should think about AI similarly to hybrid cloud strategies. Serious enterprises don’t put all their infrastructure into a single provider; they diversify to mitigate risk. The same principle applies to AI models. No single model will serve every use case optimally. Performance, cost, and specialization vary widely.

Right now, pricing is relatively low because the market is heavily subsidized. But as usage increases, costs will rise, and predictability may diminish. That is why portability and flexibility are critical. Beyond data sovereignty, organizations must begin thinking about intelligence sovereignty, maintaining control over how intelligence is developed, deployed, and managed within their own ecosystem.

Rethinking the “Best Model” Narrative

Krishna Baidya: There’s constant discussion about which model is “best.” How should organizations approach model selection?

Tejas Gadhia: The idea of a single “best” model is misleading. Most people cannot clearly articulate why one model is superior; they rely on benchmarks or headlines. In reality, most use cases do not require the largest foundation models available.

Specialized models often perform specific tasks more efficiently and cost-effectively. Large foundation models are excellent for broad capabilities and pushing boundaries, but as efficiency becomes more important, smaller and more focused models will gain traction. Eventually, organizations will likely train models on their own data to maintain control and precision.

The future will not be dominated by one model but by a hybrid approach, optimized for performance, cost, and purpose.

Native AI and Vertical Integration

Krishna Baidya: Zoho has emphasized vertical integration and native AI. How does that strategy differentiate your approach?

Tejas Gadhia: Zoho is truly vertically integrated. We build and manage our data centers, infrastructure, software stack, firewalls, and internal AI models. This was initially driven by cost control. When you depend heavily on third-party vendors, you lose pricing predictability and are forced to pass costs to customers.

By owning the stack, we maintain stability. But vertical integration also enhances context. AI’s effectiveness depends on context, understanding customer data, employee interactions, collaboration histories, and workflow patterns. When everything exists under one unified ecosystem, context flows naturally without excessive integration effort.

We provide two options: internally built AI models included at no additional cost in subscriptions, and the flexibility to connect external models through bring-your-own-key integrations. This ensures accessibility, flexibility, and sovereignty. Owning the stack enables us to optimize across hardware, models, and software layers, delivering deeper integration rather than superficial add-ons.

Templates, Extensibility, and Customization

Krishna Baidya: Many AI tools appear simple: type a prompt, get an output. How should organizations think about extending these features meaningfully?

Tejas Gadhia: Generic AI features are essentially blueprints. True value emerges when organizations customize them. For example, a note summary feature is only powerful when tailored to what a specific role or department actually needs.

Vendors can provide templates and starting points, but customization is essential. That’s why APIs (Application Programming Interfaces) and extensibility matter. Organizations must adapt AI tools to their own context rather than relying solely on off-the-shelf features.

Context is difficult to articulate manually. The nuances of workflows and processes often require ambient intelligence that continuously learns from an organization’s own data. That is far more effective than outsourcing context to external services.

The Governance Console of the Future

Krishna Baidya: How do you see enterprises managing multiple models, costs, and governance requirements going forward?

Tejas Gadhia: No organization will adopt a single-model strategy. Different models will serve different use cases with varying cost structures and performance trade-offs. Enterprises will need centralized oversight to manage experimentation, optimize costs, monitor compliance, and conduct A/B testing.

For example, a high-cost model may deliver slightly higher accuracy, but a lower-cost model might be sufficient for most use cases. Organizations need transparency to evaluate those trade-offs effectively.

A centralized governance console that integrates logging, cost monitoring, security controls, and optimization will become essential. These dimensions are interconnected and must be managed holistically.

Vibe Coding and Enterprise Reality

Krishna Baidya: There’s excitement around AI-driven “vibe coding.” Is it truly as easy as it appears?

Tejas Gadhia: The internet often exaggerates what is possible. While AI can help create quick prototypes, enterprise-grade systems require more than visual appeal. They must be secure, production-ready, governed, and maintainable over time.

Migration costs may decrease as AI tools simplify transitions between vendors. When portability becomes easy, trust becomes the deciding factor. Organizations will remain with vendors not because switching is difficult, but because they trust consistent value delivery.

For microbusinesses, vibe coding offers tremendous opportunity. For larger enterprises, however, governance and longevity remain non-negotiable.

The Human Premium in an Automated World

Krishna Baidya: With automation advancing rapidly, how should organizations preserve the human element?

Tejas Gadhia: Automation should not eliminate people; it should empower them. If bots begin communicating exclusively with other bots, the human connection disappears. That creates a transactional, impersonal experience.

Human interaction will remain valuable, perhaps even commanding a premium. When people feel empowered to build, experiment, and contribute beyond rigid boundaries, innovation compounds. Creativity and technical rigor must work together. The goal is to remove repetitive burdens so people can focus on meaningful contributions.

Start Small, Compound Value

Krishna Baidya: What practical advice would you give organizations beginning their AI journey?

Tejas Gadhia: Start small and iterate. Large, overly ambitious projects often stall. Focus on incremental improvements: routing tickets more efficiently, improving document access, or simplifying internal workflows. These may not be flashy, but they generate real value.

Each success compounds into larger innovation. Experimentation fuels creativity, and creativity fuels transformation.

Trust as the Foundation

Krishna Baidya: How would you want audiences to remember Zoho in this AI-driven era?

Tejas Gadhia: Trust is the most underestimated factor. Organizations are not just buying software; they are forming partnerships. They must evaluate whether a vendor has a track record of driving value, controlling costs, and avoiding exploitative lock-ins.

Zoho’s private structure allows us to focus on customers and employees rather than quarterly pressures. Our 30-year journey reflects patience, incremental value creation, and long-term thinking. AI is important, but trust, stability, and partnership are what endure.

Closing Reflection

As AI reshapes the business landscape, transformation will not be defined by the flashiest models or the boldest claims. It will be shaped by grounded experimentation, strong governance, vertical integration, and human-centered empowerment.

Zoho’s strategy illustrates that intelligence alone is not enough. Sustainable growth requires trust, context, and long-term commitment. In an era of rapid automation, the most enduring competitive advantage may lie not in replacing people, but in enabling them to build, create, and lead with confidence.