Don’t Skip Ahead: Why Information Readiness Comes Before AI
It’s tempting to begin AI planning with tools, models, and potential use cases. That approach makes sense from a technology perspective, but it overlooks the foundation a functional AI strategy depends on, the organization’s information.
A model or platform can provide the capabilities, but the information behind it helps determine the quality and reliability of what it produces. If that information lacks context, contains outdated or duplicate content, or has unclear ownership, those issues carry into AI initiatives, making all that implementation work less successful.
A stronger starting point is the information those systems will rely on. Before scaling AI, organizations need to have a strong grasp on their information. That foundation helps organizations make more informed decisions about where AI can add value, what information it should access, and how to manage the risks that come with its use.
As Access’s Jennifer Crawford put it in a recent webinar, “AI can only be as useful, reliable, and defensible as the information it’s built on.”
Before investing further in AI, your organization needs to understand what information it has, who owns it, how long it should be kept, whether it can be trusted, and whether it should be exposed to AI at all.
AI engineer Chris Tate says that means records, data, and information governance teams should be at the center of AI planning: “At the core of every AI strategy, every AI initiative, every AI project, is data.”
AI can expose information problems that organizations have managed around for years, like a shared drive containing multiple versions of the same document, a legacy system that’s home to records that should have been deleted long ago, or variations in metadata between departments.
AI does not eliminate those issues. It can make them more visible and potentially extend their impact across the organization. Information readiness gives organizations a way to address those conditions before they scale AI.
Before expanding AI use, build an honest picture of your organization’s information.
That starts with inventory. Your organization should know where information lives, what it contains, who owns it, how teams use it, how long it should be retained, and who should have access to it.
As you approach this work, Tate says, “assume that you don’t know where your stuff is.” Your information is most likely spread across shared drives, cloud repositories, email, collaboration platforms, databases, paper records, archives, and other locations that each team manages differently.
Finding information represents only one part of the challenge. A file can exist and remain searchable while containing outdated information, duplicate content, or material that an AI system should not access.
This inventory-first mindset can help organizations uncover information gaps before those gaps affect an AI initiative. An inventory can also reveal opportunities to consolidate repositories, improve metadata, address access issues, and establish clearer information policies.
After assessing inventory, it’s likely you’ve uncovered some redundant, outdated, and trivial (ROT) content living in legacy locations. This data can affect search results, make it harder for employees to determine which information to trust, and increase security and compliance risks.
However, this doesn’t mean your organization needs to clean up every legacy record collection before starting an AI initiative. Tate advises organizations to separate legacy remediation projects from go-forward practices and consider where practical business needs may deviate from the ideal environment.
Start by cleaning up older content that creates the greatest risk or could significantly affect an AI use case. At the same time, establish stronger practices for information created today. Those practices might address clearer ownership, better metadata, defined retention requirements, appropriate access controls, and consistent naming or classification standards.
This approach will allow your organization to make progress without treating information cleanup as a single, all-or-nothing project. Teams can address the most important legacy issues while preventing new information from creating the same problems.
AI governance needs to function as an active business practice— creating a policy and saving it within a document repository is not enough.
Leadership, legal, technology, security, business teams, and information governance professionals all have a role in determining appropriate AI use. Together, they can establish boundaries around sensitive information, approved tools, access, retention, security, and human review.
Human oversight needs a place in the process, or as Tate says, “human first, AI second.” AI may summarize information, classify content, recommend actions, or route work, but people should remain accountable for sensitive or high-impact decisions. Employees need to understand when they can rely on AI output, when they need to verify it, and when they should escalate an issue.
Governance should evolve as the organization learns. New AI capabilities, changing regulations, emerging risks, and lessons from early projects can all affect existing guidelines. Regular review keeps governance connected to actual business practices.
AI readiness doesn’t develop overnight. Starting with one or two lower-risk AI use cases gives teams a practical way to test the quality of their information, evaluate governance practices, review processes, and determine measures for success while identifying gaps that need attention.
Importantly, education should be part of these projects. Employees need clear guidance on approved tools, appropriate information use, output validation, and when to escalate concerns. That guidance can build confidence while helping employees avoid both unnecessary hesitation and overly casual use of AI.
This guiding approach can also help soothe worry about technology replacing or changing employee roles, according to Tate: “the best way to address fear is through education.”
Moving forward, build that education into existing training programs, onboarding, technology rollouts, and information governance initiatives. Then, each new AI project becomes an opportunity to strengthen organizational knowledge and improve the next implementation.
AI readiness does not require your organization to resolve every legacy information problem before it begins experimenting, but tt does require a clear understanding of the gaps that could affect AI effectiveness. Build governance into AI initiatives from the beginning, use small projects to test deliberate risk decisions, and use what you learn to inform future projects.
Information readiness gives your organization a stronger foundation for whatever comes next, improving the way you manage data beyond AI and creating clearer practices for retention, access, security, discovery, and lifecycle management.
For a deeper look at the relationship between AI and information readiness, watch the full webinar, “Chutes and Ladders: Why Information Readiness Comes First,” which is part of our series on AI and its impact on information management.
Share