4 critical questions to answer before adopting AI in your business

4 Questions to Ask Yourself Before Betting on AI in Your Business

AI offers a chance to make information more accessible and boost productivity. Yet, business leaders at Snowflake Summit 2025 highlighted that certain foundational elements must be in place before fully committing to AI.

1. What's My Cloud Strategy?

Wayne Filin-Matthews, chief enterprise architect at AstraZeneca, shared how the company is advancing AI in both research and commercial areas. For example, they've created an AI research assistant to help scientists focus on reproducibility and drug development. They also collaborate with institutions like Stanford University to experiment with agentic AI—AI systems that act as team members supporting traditional researchers.

On the commercial side, AstraZeneca leverages AI to automate content creation for marketing across 126 markets, simplifying a complex process. Yet, these efforts depend on strong data foundations. Filin-Matthews emphasized that businesses can't prioritize AI without first having a solid cloud infrastructure. "You cannot be AI-first without being cloud-first," he said.

2. Have I Addressed Data Governance Concerns?

Amit Patel, chief data officer at Truist Wholesale Banking, learned critical lessons from rolling out AI. First, banks must prove their data's source, quality, governance, and lineage to regulators. This means AI models can only access governed, authorized data sources, not just any internal data. This requirement often exposes gaps in data reliability that need fixing before AI can be safely deployed.

Second, Patel noted that many assume deploying large language models (LLMs) in a company is as simple as using AI at home. It isn’t. Defining guardrails and metadata to guide these models takes time and care. His team manages expectations by clarifying that AI implementation is not a quick, point-and-click process but requires thoughtful governance and structure.

3. What's the Quality of My Outputs?

Anahita Tafvizi, chief data and analytics officer at Snowflake, highlighted the challenge of balancing innovation speed with governance. Her team develops AI-enabled products and also uses them internally. For instance, they helped build an AI assistant for Snowflake’s sales team using Snowflake Intelligence, launched at the Summit.

Quality is key. Tafvizi raised the question: Is 95% accuracy good enough? Trust in AI outputs depends on strong governance, access controls, lineage tracking, metadata, and semantic models. These elements help maintain quality while moving fast, ensuring staff confidence in AI-generated results.

4. Have I Considered Unanticipated Benefits?

Thomas Bodenski, chief data and analytics officer at TS Imagine, shared how AI has reduced employee workloads since late 2023. But the advantages go beyond just cutting effort. AI enables tasks to be done faster, better, and with greater coverage.

For example, TS Imagine receives about 100,000 emails yearly from data vendors about product changes. Reading and processing these emails used to require two and a half full-time employees. Missing details could cause serious system failures affecting thousands of traders and risk managers.

Now, using Snowflake’s AI models, the company never misses critical information. The staff previously doing manual data curation can focus on higher-value knowledge work. AI also helps handle customer requests on Saturdays, a day with no staff coverage, by automatically responding and routing inquiries.

Before betting on AI, assess your cloud readiness, data governance, output quality, and keep an eye on unexpected benefits. These four questions can help ensure your AI initiatives deliver real value and avoid common pitfalls.