STATE A CLAIM.
Say what could be true. Avoid building a machine that only agrees with you.
THE INTERNET DOESN'T NEED ANOTHER CLAIM TO FAME.
Take a question. Connect it to real creative work. Turn the collision into a small, falsifiable experiment. Bring in ChatGPT as a collaborator—not a substitute for judgment.
One useful output, not 30 empty buttons. Type something you believe, suspect, or want to build. The machine finds related public projects and produces a critical working brief.
What this engine actually does: transparent rule-based matching and structured prompts. It does not secretly call ChatGPT, invent study results, or claim causal evidence.
Generate a brief to get a testable question, an alternative explanation, a concrete test, and three pathways into the Sahilverse.
The first brief is generated locally by transparent rules. Want a deeper critique? Copy a structured prompt with sources and limitations into ChatGPT. Nothing is transmitted until you choose to paste it.
Explore the actual materials that feed the machine. These links are catalogue entries, not guarantees that every site is currently online.
Checking whether the Live Atlas feed is reachable.
Loading publicly listed projects…
This index is maintained from public catalogue descriptions. These addresses have not all been individually checked for uptime.
Only saved in this browser. No account, analytics collection, or cloud synchronisation.
No saved briefs yet.
Something being made with an AI tool doesn't make it good, bad, original, or derivative by itself. Those are questions for evidence, context, attribution, and craft.
Say what could be true. Avoid building a machine that only agrees with you.
A plausible competing explanation is the beginning of better judgment.
Choose an observable signal before you're seduced by the outcome.
Link to sources, credit tools, mark speculation, acknowledge limitations.