M
Meaning
What real problem are we solving?
If we can't name the problem in one sentence, we don't start. Most AI projects fail because nobody took the time to frame what was actually being solved.
METHOD · THE MAKIA FILTER
The MAKIA framework, explained
Five questions, asked before any AI deployment. The filter rules out 80 % of 'AI-for-AI's-sake' projects and keeps the 20 % that actually create value.
A
Actors
Who will use it, and who decides?
An AI project without a named user is a comms project. We name the end user, the decision-maker and the sponsor before touching a single tool.
K
Knowledge
What data, what expertise?
Models know nothing about your business. We map what must be injected (documents, knowledge base, human expertise) and what stays out of scope.
I
Impact
How do we measure the value created?
If we can't say 'this will be worth X minutes a day or X €' up front, we won't measure anything afterwards. We define an impact metric at scoping.
A
Alignment
Aligned with culture & GDPR?
GDPR, data sovereignty, and internal culture. If a use case doesn't pass Alignment, we drop it — even if it's technically appealing.
This filter eliminates 80% of 'AI-for-AI's-sake' projects. We keep the 20% that actually create value.