- Most business AI runs on documents and records you already have, not on big data.
- Check five things: location, format, quality, access rights and ownership.
- Fix data problems inside the project plan, one workflow at a time.
Business leaders often assume AI needs a data warehouse and a team of data scientists. For most workflow automation it does not. It needs the right documents and records, in a usable state, that the tool is allowed to read.
The five-point checklist
1. Location
Do you know where the information for this process lives? Shared drives, email folders, an ERP, a CRM, paper files? List them.
2. Format
Is it digital and machine-readable? Scanned PDFs and photos can be read by modern AI, but typed documents and structured records are more reliable.
3. Quality
Is it current and consistent? Three versions of the same price list, or customer records without email addresses, will produce wrong answers.
4. Access
Can the AI tool legally and technically reach it? Check permissions, system connections, and whether personal data may be processed for this purpose under the Personal Data (Privacy) Ordinance or your local rules.
5. Ownership
Who keeps it up to date? An internal knowledge assistant built on last year's policies is worse than none.
Score and decide
If a workflow passes all five, start building. If it fails one or two, include the fix in the project: for example, consolidating the price list or cleaning the customer database before automating quotations. If it fails three or more, pick a different first project.
Data work is implementation work
Cleaning and connecting data is often half the effort of an AI project. Budget for it, and make sure the person leading implementation has both the business understanding and the technical skill to do it.
Want this working in your company?
Tell us how your team works today. Within two business days you get a recommended hiring model and a scoped proposal.
Talk to us