TypeSafe AI Components

Pick which of your options fits the content best (a Jev choice question). The answer is always one of the option keys you supply, with a probability per option and a confidence score, so it can drive a Switch or Condition directly. Asking several things about the same content? Use Ask Questions instead: one request is faster and cheaper than chaining components, because the content is sent and billed only once. Works best with English content.

Ask several questions about the same content in one TypeSafe Jev request, mixing all three types: choice (Classify Text), score (Score Text) and noul (Verify Statement). This is the recommended way to ask more than one thing: the content is sent and billed once and the questions are evaluated in parallel, so it is faster and cheaper than one component per question. The answers are returned keyed by your question ids. Works best with English content.

Retrieves the TypeSafe models available to your account, with their names, descriptions and release dates.

Rate the content on your own scale of 2 to 10 levels (a Jev score question). Returns the probability-weighted score, the most likely level with its description, a probability per level and a confidence score. Asking several things about the same content? Use Ask Questions instead: one request is faster and cheaper than chaining components, because the content is sent and billed only once. Works best with English content.

Performs an arbitrary authorized call to the TypeSafe API, for example POST /v1/systemone with a request you build yourself.

Check whether a statement is true for the content (a Jev noul question). Returns the probability that the statement holds and a true/false result against your threshold. Asking several things about the same content? Use Ask Questions instead: one request is faster and cheaper than chaining components, because the content is sent and billed only once. Works best with English content.