You need a data plan
If you’re focusing your resources on building your agentic layer and your plan is to just connect it to your existing data as-is, you’re doing it wrong.
Does your agent know where the data is and how to access it with the right permissions? Your data is sitting across multiple data platforms (Databricks, Snowflake, etc.), hyperscalers (Amazon Web Services (AWS), Google Cloud, Microsoft Azure, etc.), and enterprise systems ( Microsoft 365, Workday, Adobe Workfront, etc.). You can federate it, but is the permission model uniformly applied? Iceberg is increasingly dissolving data silo barriers, but agents will need a single source of permission truth.
Does your agent know what the data means? Context is critical to ensure your agents know what the data is and how to use it: semantics (how it’s defined), ontology (how it relates to everything else in your business), taxonomy (how it’s classified and organized), and more. Recency, frequency, and relevancy are key contextual signals agents need to ensure they’re working with the right data at the right time.
And the idea that you’ll have one agentic layer across your enterprise that becomes the single keeper of knowledge for your company is, IMHO, legacy thinking anchoring you to the way things used to be. AI is a foundational shift equal to, and possibly even surpassing, “the cloud.” Software systems of the future won’t just be AI-enabled. When RSI (recursive self-improvement) fully arrives, it’ll all be AI-driven, and all areas of your enterprise will evolve toward autonomous agent-to-agent workflows that build, monitor, and improve themselves. It will be bumpy, but it will get there, and given the pace of innovation, sooner than we think.
Every one of those autonomous agents will need data it can find, access, and understand. Without data readiness, they’ll use, share, and iterate on different data because there’s no single source of truth.
My point is: if you don’t have a strategy, approach, and plan for your data, you need one, starting now.