三個問題揭示企業能否掌控自身的 AI

4 people in an open office, at computers

By Mahmood Majeed, global digital and technology leader at ZS

AI is becoming business infrastructure, but many companies still don't control it.

That control depends on the choices companies retain, the business context they own, and the accountability they establish for results.

You can test all three by asking these strategic questions: How much dependency on AI providers is too much? What makes your AI yours? And do you control the value AI delivers?

How much dependency on AI providers is too much?

Nearly six in 10 pharmaceutical and biotechnology companies expect to rely on major AI providers in 2027, according to our latest outlook survey of technology leaders and chief data officers. That dependence leaves companies exposed to vendor pricing changes and costs that become harder to predict as usage grows. Other risks include outages, vendor lock-in, and regulatory concerns. When AI sovereignty is a factor, geopolitical risk also comes into play.

Staying in control means preserving the flexibility to shift if needed, including changing providers and maintaining appropriate leverage to negotiate future agreements and keep costs in check.

Companies can minimize their dependency by keeping their core data engineering under their control, running sensitive applications in controlled environments, diversifying across proprietary and open-weight models where needed, and designing systems that can swap models without rebuilding workflows.

What makes your AI yours?

Give two pharma companies the same models and claims data, and the difference comes down to what they build in: their expertise, playbooks, and ways of working.

As agents take on more tasks, companies need AI-ready data. People need the skills to build, direct, operate, and evaluate agentic systems. Governance needs clear boundaries for autonomy, and workflows need to be redesigned around what AI makes possible. Without those capabilities, companies default to generic AI or lose control to agents that duplicate work, act unpredictably, or work at cross-purposes. They can also become too dependent on outside specialists to run their AI.

We're seeing more pharma companies control how they put their own business rules and processes at the core of their AI. They're building systems that use their enterprise data and connect to business applications to redesign workflows suited to different business contexts and areas of work.

Making your AI truly yours depends on the enterprise data, talent readiness, operating model, and governance behind it.

Do you control the value AI delivers?

CIO budgets for AI keep climbing, yet the gains remain hard to spot on the P&L.

The disconnect often starts with ownership. Who owns the business outcome, and what counts as success? Business leaders should own the result, with technology and finance providing the capabilities, investment discipline, and measurement systems to deliver it.

One finding in our survey shows how widely technology leaders recognize the problem: 70% say changing how business and technology teams collaborate is a priority for 2027, with business, technology, and finance aligned on shared KPIs. Shared measures turn progress into business terms, making it easier to decide which AI projects to champion, scale, or end.

Turning that progress into sustained business impact and competitive differentiation requires clear ownership of AI adoption, change management, and new ways of working.

These few questions can help you clarify that ownership. Your questions may differ in 2027, but the goal remains the same: Stay in control and turn AI investment into value.

Turn AI investment into measurable value with ZS.

This sponsored post was supplied by ZS.