How I research, write, check and correct these notes

§ 1 How I make a field note

I research and draft each note with AI assistance. I then review and edit it myself before it is published. I am responsible for everything on this site.

I am an AI consultant with a search marketing background. I am not a data engineer. I explain what primary sources say. I do not report my own tests, benchmarks, client work or first-hand engineering experience. One note is an exception. Prompt length vs brand mentions reports figures from my own tracking data. That data is not public. The note says so at the top. More on that on the about page.

§ 2 Sourcing

• I link every factual claim to a source. Each note ends with the full list of sources I used.
• I go to primary sources first. That means project documentation and specifications, standards bodies, vendor announcements and the original papers and essays.
• Each note states the date on which I checked its sources. Software moves quickly, so read version numbers and release dates against that date.
• If I cannot verify a claim, I cut it. Sometimes I could not read a source in full, for example a paid standard. The note says so when that happens.
• I summarise sources in my own words and limit quotation to short phrases.
• I label opinion and analysis as such.
• My diagrams are conceptual. I do not draw charts of numbers unless every number comes from a cited source. The one data chart so far, in the prompt length note, draws only the figures printed in the table above it.

§ 3 Corrections

If a note contains an error, I correct it and add a dated correction note saying what was changed. The "last updated" date on the note changes at the same time. I do not make corrections silently.

To report an error, use the contact form. Include the title of the note and a source if you can.

§ 4 Conflicts of interest

I work commercially in search marketing. I founded Code Scout, a desktop AI coding agent. So I have an interest in AI tools. When a note uses my own data, it says so at the top.

The site promotes two things. Both are my own. The first is my consulting services (AI data readiness audits, private AI implementation and training), set out on the hire page. The second is Code Scout, a free app described on its own page. Apart from those, it carries no advertising, sponsored entries or affiliate links.

§ 5 Same-owner sites

I own and write three other sites. They are separate and cover different subjects.

• frankvitetta.com: AI search and SEO
• frankvitetta.co: artificial intelligence
• frankvitetta.org: UK technology

This one, frankvitetta.net, covers big data. That means data platforms, table formats, analytics engines, data quality and measurement.