Lrnon Lrnon teaches people to use AI well — free for every learner, and without anyone paying us to recommend anything. This page is the part worth reading before you trust the rest.
Trust is the whole product.
A learning platform is worth exactly what its readers believe it. One instance of undisclosed favouritism would cost more than any commercial arrangement could pay us. Every commitment below exists to protect that, and most of them are enforced by our own software rather than left to good intentions.
No vendor pays for placement, ranking or inclusion in a lesson, a tool card or a comparison, and none can. There are no affiliate links and no paid reviews. Lrnon does take donations, grants and, under strict published rules, advertising, and none of it is allowed to touch what is taught or who is described. The rules are on the Support page.
We name real products, because teaching about AI without naming what people actually use would be useless. Naming a tool is not endorsing it. We describe what exists and teach you a method for choosing — the right answer depends on your language, budget, jurisdiction and work, and we do not know those.
Descriptions come from the vendor’s own published material, or from what the market has openly tried and reported. Every tool named in a lesson shows a link to that source and the date we last checked it.
Where our description and a vendor’s current documentation disagree, theirs is right and ours is out of date. The link is always there so you can reach the authority directly rather than relying on us.
A tool description older than 120 days fails our build, and shows you a warning banner before it gets that far. A script checks vendors’ documentation for changes. This is enforced by the software rather than promised by us.
Where a question is unsettled, the lessons say so and say what would settle it. Where a limitation may since have been fixed, they tell you to test it yourself. Manufactured certainty is the failure we work hardest to avoid.
Some courses avoid naming tools, on the theory that it looks impartial. We think that fails the reader twice: it makes the teaching abstract, and it hides real differences behind a pretence of even-handedness.
Our own module on verification argues that presenting a lopsided situation as an even one misrepresents the state of knowledge. We hold ourselves to that. If one tool is genuinely better for a task, saying so with the evidence and the date is honesty, not favouritism. What we will not do is prefer anything we have not shown you the reason for — or anything anyone has paid us to prefer, which will never happen.
The choice stays yours. We describe the field and teach you how to run your own comparison, because your language, your budget and your work decide the answer, and we cannot see any of them.
AI changes faster than any curriculum can. Our answer is structural rather than aspirational: lessons teach mechanisms, which do not change when a vendor ships a release, and volatile facts live in a tool registry that is dated, checked, and fails our build when it goes stale.
Content is captured on a date, from a source we show you. If a vendor changes something afterwards, the link takes you to their current position — which is the authority. As Lrnon grows, keeping that current stays a first-order priority rather than a backlog item.
Found something out of date or wrong? Tell us — corrections are published in what's new rather than made quietly.
One person answers, within two working days: lrnon.org@gmail.com. See also privacy, terms and child safety.