“Is our data ready for AI?” is a hard question to answer for a whole business. It gets much easier when you ask it about one job.
The film walks through that check in 103.4 seconds, using a plumbing firm as a hypothetical example. The written guide below covers the same four steps in a little more detail.
Why the data matters
An AI system can only help with a job if it can reach the information that job needs, and that information is right. When either part is missing, the output can look finished and still be wrong.
In February 2025, Gartner predicted that through 2026, organizations will abandon 60% of AI projects that lack AI-ready data.
Step 1: Pick one job
Choose a single piece of work with a clear output. In the film, a plumbing firm wants AI to draft the 40 boiler quotes Sarah writes every week. The firm, Sarah, her customers and every number in the example are invented.
A narrow job gives you something to test. “Draft a boiler quote for Sarah to check” has a clear finish line, and you can compare the drafts with quotes she has already written.
Step 2: List the data the job uses
Write down every source the job draws on. For a boiler quote in the example, there are four.
| Source | Where it is | What the check finds |
|---|---|---|
| Customer emails | Shared inbox | Easy to reach. Names differ from other records, and some emails include bank details. |
| Price list | Office wall | Prices from March 2023. |
| Old quotes | Filing cabinet | Paper only. |
| Supplier prices | Trade counter | Current as of October 2026. |
Make the list with the person who does the job. They know which sources they actually use, including the ones nobody wrote down.
Step 3: Check each piece
Ask four questions about every source on the list.
- Can the AI get to it? The old quotes are on paper, so for now the answer is no.
- Is it up to date? The price list is from 2023, so every quote built on it would come out wrong.
- Is it consistent? The inbox says J. Smith and the job sheets say Smith, John. A person knows that’s one customer. The AI only knows if the records are matched.
- Is it safe to use? Some customer emails include bank details. Decide what the AI is allowed to see before you connect anything.
Step 4: Score it and fix the red first
Give each source a color. Green is ready. Amber needs a tidy. Red is real work.
Start with the red items, because they block the job. In the example, that’s the paper quotes. Turning paper records into usable data is work AI does well, with a person checking what it extracts.
Where it gets deeper
Each check has more to it. Matching customer records reliably, with a score for each match and a person reviewing the uncertain ones, is its own piece of engineering. So is making sure the AI only retrieves what each user is allowed to see, and testing and monitoring the system once it runs.
That’s the work Hone does. We’re AI engineers who find where AI fits in your business, then build it and keep it running.
Sources
- Gartner, “Lack of AI-Ready Data Puts AI Projects at Risk”, February 26, 2025.
Want to check the data behind a job in your business? Book 30 minutes with us and bring the job you have in mind.