Our documents are a mess. Clean them up before AI, or after?
Gartner expects 60% of AI projects to be abandoned due to lacking metadata management and data quality. Without a unified data foundation, your AI initiatives risk failure and inconsistent experiences.
The question
Our internal documentation is inconsistent, outdated in places, and was never structured for retrieval. Do we invest in cleaning and structuring it before deploying AI over it, deploy first and let real usage reveal what needs fixing, or scope the cleanup narrowly to the one workflow we are trying to prove?
Counsel's position
Scope cleanup narrowly to a critical workflow, ensuring high data quality for that segment before AI deployment.
Verdict
The verdict: Scope cleanup narrowly to a critical workflow, ensuring high data quality for that segment before AI deployment.
How the criteria decide
4 of 5 criteria resolved on cited evidence. 1 had none either way.
| Criterion | Favours | Evidence |
|---|---|---|
| data quality | Scope cleanup narrowly | AI projects will be abandoned due to poor data readiness Saying whether or not something passed your schema makes your downstream systems much more reliable. Only enterprises consider their data completely ready for AI It behooves these two disciplines to share a definition of “done” that includes the data being ready before the model is deployed rather than after it fails. Point-tool AI workflows fail without a unified data foundation Without a unified, well-governed data foundation, they lack industry and organizational context, are siloed from key enterprise systems, are only built for reading, not activating. |
| user experience | Scope cleanup narrowly | Scoping AI knowledge apps to a single recurring workflow validates impact Start small. Pick one corner of your team's knowledge that people keep asking about, stand up an application for it, see if it changes the shape of your week. |
| time to deploy | Scope cleanup narrowly | Scoping AI knowledge apps to a single recurring workflow validates impact Start small. Pick one corner of your team's knowledge that people keep asking about, stand up an application for it, see if it changes the shape of your week. |
| resource cost | No evidence either way | |
| workflow impact | Scope cleanup narrowly | Scoping AI knowledge apps to a single recurring workflow validates impact Start small. Pick one corner of your team's knowledge that people keep asking about, stand up an application for it, see if it changes the shape of your week. |
AI projects will be abandoned due to poor data readiness
Relying on post-collection data cleaning risks model inaccuracy; capturing metadata and provenance at the source is required before deploying AI over your documentation.
Only enterprises consider their data completely ready for AI
Data quality and readiness are the top obstacles to AI success, meaning your inconsistent documentation must be addressed before model deployment, not after.
Point-tool AI workflows fail without a unified data foundation
Building agent workflows on fragmented, unstructured documents yields inconsistent experiences that are impossible to scale.
AI agents require a semantic context layer to reason over data
Unstructured documentation lacks the attached meaning, entity relationships, and temporal validity that agents need to function autonomously at runtime.
Scoping AI knowledge apps to a single recurring workflow validates impact
Starting small with one corner of your team's knowledge allows you to stand up a reliable application without extensive engineering overhead.
Read another verdict
- Start with a small test, or take on the whole process at once?
- Our competitors advertise AI and we don't — match them, or hold the line?
- Our people already put client files into ChatGPT — ban it, frame it, or supply a tool?
- Our most experienced person retires in two years — how do we keep what they know?
- We can't hire the experienced people we need — automate, train up, or outsource?
- Slow our EU AI Act prep now the deadline's moved to 2027?
- Use AI to flatten middle management this year?
- Let an AI agent act on its own — or keep a human in the loop?