SYSTEMONEBuilt by Ordinexis Decision playground
Powered by Cloudflare Clef
LANGUAGE/EXPERIMENT 01

Answer from a source

A

Source content

B

Questions & answers

1 question

Live runs send your inputs to Cloudflare Workers AI. This app does not store them.

Decision output

READY TO TEST
Request & response JSON

After a run, the request shows the inputs used for the displayed result. Image data is omitted here; exports include it.

TRY A VARIATION

Choose an optionYes / noScore a rubric
A DIFFERENT KIND OF MODEL

A decision, with a distribution.

System One models evaluate evidence against a schema you define. Clef selects among your allowed answers and returns probabilities. It does not write an open-ended chat response.

  1. Content A is the evidence. Paste text or JSON, or attach images.
  2. Content B defines the decision. Write a question, then select yes/no, answer options, or an ordered scoring rubric.
  3. Run and inspect. Compare answers, probability distributions, response time, and token use.

A score is the probability-weighted average of levels starting at 0. Confidence expresses the model’s distribution, not a guarantee that its answer is correct. Add an “unknown” option when evidence could be missing.

Example outputs are hand-authored illustrations for preset inputs. They are not measured model results. Latency and token use appear only after a live run.