Skip to content
Jev Hub

[ Use case · Data ]

Can Jev sort documents into a long list of categories?

Pick one of dozens of categories, and fall back to a broader one when unsure.

Published By Jev Hub

[ TLDR ]Good fit
  • Yes, and its confidence tells you which answers to trust.
  • In TypeSafe's cookbook, 60 company annual reports were classified into 75 industry groups with one Choice question each.
  • A confidence cutoff of 0.9 split them in half: the confident half was right 90% of the time, the other half 40%, and reporting that half one level up the hierarchy raised it to 70%.

Is Jev a fit for documents?

Use Jev when:

  • Your categories form a hierarchy, so an unsure answer can fall back to the broader group.
  • There are many categories (Jev's Choice takes up to 255 options).
  • A broad but right label is more useful than a precise but wrong one.

Skip it when:

  • Every document must get a precise label with no review. Plan for the unsure half.
  • Your labels are not described anywhere. Each option needs words that describe what belongs in it.

Which questions does Jev answer?

One request per document, one question. Jev answers them in parallel.

  • Which industry group best describes the business this company runs today? (choice. Options: Depository institutions, Insurance carriers, Business services, Chemicals)
ONE DOCUMENTJEV ANSWERSWhich category?YOUR CODEConfident: keep itUnsure: go broader
One document, one question, then your code. Jev answers the questions about each document in one request. Your own code reads the answers and their probabilities and decides what happens next.

What does it cost?

About $1.13 a month for 10,000 documents of about 1130 words each: 11¢ per 1,000 at list price, billed by TypeSafe AI. Adjust the numbers in the calculator.

What does TypeSafe's worked example show?

TypeSafe's cookbook classifies 60 SEC annual reports (10-K) into the 75 industry groups of the Standard Industrial Classification, which roll up into ten broader divisions.

  • A confidence cutoff of 0.9 split the filings in half.
  • The confident half was right 90% of the time; the other half 40%.
  • Reported one level up as a division, that uncertain half was right 70% of the time, with no extra request.
SHARE CORRECTConfident half90%Unsure half40%Unsure, one level up70%
Confidence separates the answers to trust. Share correct for 60 annual reports in TypeSafe's own cookbook, split at a confidence of 0.9.

Source: TypeSafe's cookbook Classification using confidence (opens in a new tab), numbers from jev-1.12 on 12 August 2026. These are the maker's own results on its chosen data; test on yours.

How do you get started?

  1. Test it in the playground. No code needed. Paste a real document into TypeSafe's playground and ask: “Which industry group best describes the business this company runs today?” with the options Depository institutions, Insurance carriers, Business services, Chemicals. You'll see the answer and its probability. Open the playground (opens in a new tab).
  2. Use an existing tool. We haven't found a ready-made tool for this workload yet. Browse what people have built with Jev.
  3. Integrate it. To run it on every document automatically, call Jev from the system where your documents live. The developer details below include the full request.

Common questions

How do developers set it up?

The request

One request per document with all 1 questions batched over the same state, so Jev reads the state once. Keys below are the names you'll see in the response.

POST api.typesafe.ai/v1/systemone
{
  "model": "jev-latest",
  "state": "Item 1. Business. We operate 140 branches across three states, taking deposits and making commercial and residential real-estate loans...",
  "questions": {
    "group": {
      "type": "choice",
      "instructions": "Which industry group best describes the business this company runs today?",
      "criteria": {
        "depository_institutions": "Banks, savings institutions and credit unions",
        "insurance_carriers": "Life, health, property and casualty insurers",
        "business_services": "Software, advertising, staffing and other services to businesses",
        "chemicals": "Chemicals, plastics, drugs and allied products"
      }
    }
  }
}

How to wire it up

  1. Describe each category by what falls inside it, not only by its name.
  2. Send one document with one Choice question over all the categories.
  3. Keep the label when confidence is high (the cookbook uses 0.9).
  4. Below the cutoff, report the broader parent category instead; it follows from the same answer, so there is no second call.

Watch out for

  • Hard documents look like easy ones from the label alone. Read the confidence, not just the winning option.
  • Send the part of the document that describes the business, not the whole filing.

New to writing Jev questions? Read the question design guide.