Work Story
Reading AI Guidelines as a System
The Assignment
From November 2025 to January 2026, I supported research related to public-sector AI guidelines.
The task was not to invent a new policy. My role was to collect, compare, and organize existing guidance materials from public sources.
The intended users were people preparing proposals and project documents. They needed to know which source was relevant to a particular question and where to look next.
Why Collection Was Not Enough
A folder full of documents is not yet a useful guide.
Different documents can use similar words for different purposes. One source may focus on security, another on procurement, another on project management, and another on the use of AI itself.
I had to read for structure:
- who published the material;
- what kind of work it applied to;
- which question it helped answer;
- where two sources overlapped;
- where the reader still needed to check the original document.
What I Practiced
This work strengthened a different part of my technical ability.
I practiced comparing definitions, separating evidence from interpretation, and leaving a path back to the source. I also learned that documentation has an interface. Headings, categories, references, and exceptions decide whether another person can actually use the research.
What I Do Not Claim
I did not author national policy, approve an official guideline, or own the final publication. I supported research and organization within a short project period.
That distinction matters. Clear boundaries make the real contribution easier to trust.
What Stayed With Me
The same habit appears in development work. Before connecting systems, I need to know which source is authoritative, which field means what, and where an assumption entered the flow.