From AI projects to an operational capability

2026-02-19 · Source: Databricks · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Data Science & Analytics · Depth: Intermediate, medium

Summary

Databricks CIO Naveen Zutshi discusses the transition of AI from experimental projects to core operational capabilities within enterprises, emphasizing its impact on P&L and business KPIs. He notes a significant shift over the last 6-12 months, with regulated industries adopting AI for back-office automation, fraud detection, and drug discovery. Funding for AI has moved from innovation budgets to major line items, signaling operational commitment. Zutshi identifies legacy systems, SaaS sprawl, and architectural complexity as the primary bottlenecks to scaling AI, rather than talent shortages. He advocates for modern, open architectures with unified governance, bringing models to data, and avoiding single model provider lock-in. Critical platform decisions include combining structured and unstructured data under common governance and treating AI as a core capability with robust observability, quality, validation, and testing.

Key takeaway

For CIOs and business leaders navigating enterprise AI adoption, prioritize architectural modernization and unified data governance over incremental AI projects. Your focus should be on integrating AI into core business KPIs and P&L, ensuring robust testing and observability. This approach will enable scalable AI solutions and attract top engineering talent, transforming AI from a cost center to a value driver.

Key insights

AI becomes an operational capability when it impacts P&L, business KPIs, and is supported by modern architecture and governance.

Principles

• Modernize legacy systems to scale AI.
• Bring models to data, not data to models.
• Treat AI as a core capability with disciplined testing.

Method

Implement a "start, stop, continue" framework: stop feeding legacy, stop treating governance as an afterthought, and stop agent sprawl. Start with skill-based tasks and continue investing in data and governance.

In practice

• Use an AI gateway to abstract multiple model providers.
• Automate data entry with agents for productivity gains.
• A/B test agent recommendations against manual approaches.

Topics

• AI Operationalization
• Enterprise AI Strategy
• Data Governance
• Legacy System Modernization
• AI Architecture

Best for: VP of Engineering/Data, Executive, AI Architect, Director of AI/ML, CTO, MLOps Engineer

Related on AIssential

• Building the Operating Environment for AI Systems — Scalable AI requires a dedicated operating environment for governance, orchestration, and observability, moving beyond mere model capability.
• Architecting Data And AI In The Era Of Enterprise Intelligence: Meet Shylaja Nathan, Principal Analyst — Agentic AI demands modern data architectures that provide context, responsiveness, and semantic consistency at scale.
• The shifted operation model and MSPs in the AI Age — AI's shift to infrastructure demands new operating models and managed services focused on probabilistic intelligent assets.
• Bridging the operational AI gap — Integrated data and systems are critical for successful enterprise-wide AI adoption and operationalization.
• Operationalizing AI for Scale and Sovereignty — AI factories enable scalable, sustainable, and governed AI by prioritizing data control for secure, tailored solutions.
• The Foundations of Evaluating Traditional Organizations for AI Integration — Successful AI integration in traditional organizations hinges on a thorough preliminary assessment of internal readiness.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Databricks.