California is moving fast on AI adoption, but the state's technology leaders say the real work starts with the data those tools will depend on. The state has signed a partnership giving all agencies, cities, and counties access to Anthropic's Claude at 50% off, is building a statewide AI assistant called Poppy, and has begun adding AI-focused software providers to its statewide license program. None of that matters, however, if the underlying data isn't ready.
The state's push reflects a broader reality for government agencies: AI tools are only as good as the information they're fed. California's Department of Technology has been assessing how AI can help state departments deliver more effective services, and the investments are substantial. But officials and integrators working on these projects say agencies that skip data preparation will face a harder, riskier, and more expensive path to implementation.
Why data readiness comes first
Legacy systems remain a significant obstacle across state government. Recent California IT initiatives have highlighted the importance of understanding these systems and the data they contain. AI doesn't fix legacy data problems - it's actually more dependent on back-end data than standard implementations.
"Implementation of AI requires data readiness as a core component of success and the cornerstone of an AI-driven project," according to Visionary Integration Professionals (VIP), a systems integrator that has worked with more than 1,200 public sector and commercial customers since 1996, including the California Department of Consumer Affairs.
Three building blocks for AI-ready data
VIP identifies three specific areas agencies need to address before deploying AI tools.
Data standardization. Consistent formats, fields, and definitions must exist across systems so AI tools can understand the information they receive. This means taking inventory of systems, mapping them out, and identifying fields that have different names across different systems. "AI will not understand your tribal knowledge," VIP notes. Standardization isn't a one-time project - it's ongoing, as systems and data sources continue to change.
Data governance and management. Before any tool touches agency data, there need to be clear written rules about what's AI-exposable versus restricted. Governance is about decisions and accountability, not just policy documents. For agencies carrying legal exposure or working under federal regulations, this step is especially critical. Legal teams, program leads, and privacy officers all need to be involved in creating the framework.
Data access and control. Governance sets the rules; access control enforces them. This includes audit logging, encryption in transit and at rest, and role-based permissions. AI tools need to meet the same security standards as existing systems before they're connected. "Without this layer, governance is just a document with rules," VIP said.
These building blocks apply broadly across government AI work. For those looking to build skills in this area, AI for Government training can help teams understand what AI adoption requires. Similarly, Data Analysis coursework supports the core competency of preparing and understanding agency data.
California's approach sets a precedent
California's commitment to AI is notable not just for its scale, but for what it signals to other states. The state is pushing to ensure that all future IT spending takes full advantage of what AI can bring. That includes incorporating AI into state-sponsored and approved IT projects going forward.
"California is setting the tone and is a trailblazer in AI and many other states are taking notice," VIP said. The state's approach to legacy data will be watched closely as a model for how large governments can transition to AI-backed solutions.
Why this matters for government professionals
For those working in state and local government, the takeaway is practical: AI adoption in your agency will likely be shaped by data readiness long before any tool is deployed. Understanding your agency's data standards, governance rules, and access controls isn't just an IT concern - it's a job requirement for anyone involved in AI projects. The agencies that succeed will be those that treat data preparation as the first step, not an afterthought.
Your membership also unlocks:
