Understand AI. Decide where to use it.
Guides and observations to improve your brand's visibility and the way your business works. Each note separates what can be verified from what remains a hypothesis.
I publish when a source, method, or verifiable case adds something useful.
RSS feed14 articles available
[ latest article ]
[ chronological archive ]
[ 2026 ]
- [ Framework ]
How long before data becomes risky for an AI agent?
An agent is only as reliable as the data it can read. Make freshness, ownership and information limits visible.
Reading: 9 min
- [ Field guide ]
What makes a product page readable by a shopping agent
Product pages must help AI systems distinguish variants, availability and evidence, not only persuade a person to buy.
Reading: 7 min
- [ Protocol ]
How to measure visibility inside a generative engine
Useful measurement starts with declared questions, a documented protocol, and dated observations. A score alone does not prove brand visibility.
Reading: 8 min
- [ Definition ]
GEO and AEO: what actually changes in practice
GEO and AEO describe related but different surfaces. The distinction matters when the thing you need to observe changes.
Reading: 8 min
- [ Analysis ]
llms.txt: what it does, what it does not, and why I publish one anyway
llms.txt may simplify access to selected content, but it does not guarantee indexing, understanding or citation by generative engines.
Reading: 8 min
- [ Protocol ]
How to verify an AI agent's output
Delegating a task does not mean accepting its output. A validator makes evidence, criteria and stopping conditions explicit.
Reading: 8 min
- [ Field guide ]
How to help AI recognise your brand correctly
A legible brand keeps its name, role, relationships and evidence consistent across its site, profiles and independent sources.
Reading: 8 min
- [ Field guide ]
Why many AI pilots never become operating processes
A pilot becomes a process when it has an owner, stable inputs, controls and a verifiable definition of completed work.
Reading: 8 min
[ 2025 ]
- [ Technical analysis ]
Why many AI crawlers cannot see your site
What AI crawlers can read when content and links depend on JavaScript, and which checks should come before intervention.
Reading: 8 min
- [ Framework ]
What you can actually delegate to an AI agent
A task is delegable when it repeats, has accessible data and can be checked with a sufficiently reliable rule.
Reading: 8 min
- [ Technical guide ]
Structured data for AI: what helps and what is theatre
Structured data clarifies entities and relationships already present on a page. It cannot replace incomplete content or missing evidence.
Reading: 8 min
- [ Framework ]
Which business process should AI improve first?
Impact, verifiability, data access and risk help select a first AI use case without chasing the most impressive demo.
Reading: 8 min
- [ Analysis ]
How AI engines select sources for their answers
AI answers combine retrieval, selection and generation. Understanding these stages prevents confusion between ranking, presence and citation.
Reading: 9 min
How visibility inside a generative engine gets measured, written out in full. Read the article
Back home