Marketplaces · Publishers · Education teams

Block AI slop and low‑quality material
before it reaches your customers

Steletta is your AI harness that combines agents to check every question, answer, distractor and learning goal across your catalog

Find weak material.
Remove it or help creators fix it

Screen new submissions and your existing catalog, with focused checks for common AI mistakes. Give creators specific fixes before material is published or promoted

One verdict. Two scores.
Clear fixes for every resource

Send a batch in one command. Get simple, structured results your team and your software can use

One clear verdict
Ready to useorNeeds fixes
Two accurate scores
Coverage92/100
Quality72/100
Actionable feedback
Required fixes
Suggested improvements
Strengths
YOUR RESOURCES

Large batches.
One simple command

grade-6-math-review.pdfPDFreading-comprehension.docxDOCXscience-assessment.pdfPDF
npm run review:catalog -- catalog.json

The batch runner handles each upload and collects the completed reviews

See the API documentation

Receive structured results

200 OK
{
  "reviewId": "rev_example",
  "decision": "needs_fixes",
  "scores": {
    "coverage": 92,
    "accuracyQuality": 72
  },
  "findings": [
    {
      "id": "finding_01",
      "type": "required",
      "page": 2,
      "location": "Question 2",
      "title": "Two answer choices are correct",
      "evidence": "B (square) and D (rhombus) both have four equal sides. The key accepts only B",
      "suggestedAction": "Add \"and four right angles\" to make B the only correct answer"
    },
    {
      "id": "finding_02",
      "type": "required",
      "page": 4,
      "location": "Question 6 · Answer explanation",
      "title": "The explanation states a false rule",
      "evidence": "The explanation says absolute value is always positive, but the absolute value of zero is zero",
      "suggestedAction": "Explain absolute value as distance from zero, which is always nonnegative"
    },
    {
      "id": "finding_03",
      "type": "improvement",
      "page": 3,
      "location": "Question 4",
      "title": "Two equivalent choices can be ruled out",
      "evidence": "4/6 and 2/3 are equal. Because only one answer can be correct, learners can eliminate both without knowing the answer",
      "suggestedAction": "Replace one duplicate with a distinct wrong value, such as 3/8, so each choice tests a different misconception"
    },
    {
      "id": "finding_04",
      "type": "strength",
      "location": "Across the material",
      "title": "The core skills receive balanced practice",
      "evidence": "Fractions, proportional relationships and signed numbers each receive practice, with a mix of calculation and reasoning questions"
    }
  ]
}

Scroll to explore required fixes, suggested improvements and strengths

Why not just use ChatGPT?

One request.
We handle the rest

Submit your material once. No need for multiple manual API calls. Our AI harness handles the agents, multiple passes and verification for you

Your material
Steletta AI harness
ReviewRecheckVerify
One complete report

Catch subtle
AI mistakes

AI-generated material needs a separate review. Our agents target subtle flaws that can survive a model’s self-review

  • Weak distractors
  • Multiple correct answers
  • Vague questions
  • Poor coverage
  • Similar questions
  • Answer-key patterns (e.g. five Bs in a row)

Accurate, reliable
scores. Not guessed

Our proprietary formula calculates Coverage and Quality from the review evidence, using fixed rules and clear cutoffs

Needs fixes
Coverage92/100
92 out of 100
Quality72/100
72 out of 100

Let your own reviewers make the comparison. Let’s talk

Your material stays yours.
Always

We automatically delete the uploaded copy after your review

Read our privacy policy

Don’t take our word for it.
Put us to the test

Compare Steletta with your internal reviewers before deciding whether to go further

One command.
Results in minutes

  • Review your sample through the API in one command
  • Full reviews with scores and specific fixes
  • Run the pilot without connecting your live catalog
  • No technical team? We can help with setup and integration
Plan your pilot

A fair comparison.
A clear decision

  • Run the pilot
  • Compare with your own reviewers
  • Expand when the results make sense for your team