Finding public sources and recording where they come from
By Oliver Schmalzl · Written: · Archive discovery date:
20 April 2026 · Archive date
A search hit is a start, not yet evidence. A headline can shorten a claim, a preview can be out of date, and a shared post can lose its original context.
For your own note, it is therefore important where the statement actually appears and how far it goes. This applies especially to new tools, whose launch is easily mistaken for a practical recommendation.
A small research task about paper models
An invented question: which file format does a publicly offered folding template use? The right source is the publisher's page or its documentation. Someone else's summary can help you find it, but does not necessarily answer the question reliably.
The note should contain the page title, the link, the date accessed and a short summary of your own. For product documentation, the version is also of interest. Instructions for an older release should not quietly be used as a current description.
If a source only repeats a claim, it is worth looking for the original publication. If it cannot be found, that is exactly what gets recorded. Something that only sounds plausible does not yet belong in the text as a fact.
The origin stays with the result
An AI can sort suggestions and find possible sources. The finished text should still make clear which statement rests on which source. A long list of links at the end helps less than a fitting reference at the specific point.
The goal is a small, traceable answer. It may also be: "The feature has been announced; its practical quality was not tested here." This keeps the difference between a public find and your own experience visible.
Original finds
These links were the starting point for this post. The assessment in the text is mine; the linked posts are not my own tests.
- A local research setup with LM Studio
- SearXNG as your own search service
- A collection of free APIs
- Marketing tools as a public overview
- Collection of public AI projects, including n8n, LangChain and DeepSeek.
- Post about using Codex as a coding agent, with several claimed use cases.
- Andrew Ambrosino describes a major ChatGPT update with Work, a desktop app, plugins and browser features.
About the date: This post was written on 11 October 2026. The archive date matches the earliest related find in the collection, not a publication at that time. Technical assessments and sources added later refer to the time of writing.