Use Jev to rank product needs

Q: Why use Jev instead of a chat model for this?

Each need is the same three short questions over a small block of text. That is a classification job, not generation. In the source post, classifying 149 needs cost about $22.51 in Jev credit versus an estimated $5.8k with existing LLMs—roughly 250× cheaper.

Q: How do user-tunable weights fit in?

Jev labels each need once (type, severity, coverage). Your code turns those labels plus hard data—complaint count, growth, launch count—into numbers and computes a weighted score, e.g. w_growth × growth + w_severity × severity + w_gap × (1 − solved). Moving a weight re-sorts instantly with no new model calls.

Q: What does the Flatkey call look like?

POST https://router.flatkey.ai/api/alpha/decisions with Authorization: Bearer …, Content-Type: application/json, and a body of state, model (typesafe/jev-1.13), and questions (choice / score / noul). This is not an OpenAI-compatible /v1 chat endpoint.

Q: What should happen when Jev is unsure?

Low confidence usually means the need summary bundles two problems or the complaints are vague. Keep those rows out of the ranking, split or rewrite the summary, and re-ask—it is cheap enough to run twice.

Q: Can I bring my own API key?

Yes. Try the free demo above without signing in. To run it live on your own text, sign in with Google and connect a Flatkey or OpenRouter key (a quick test call runs when you save it). Manage or delete it anytime under Dashboard → BYOK (https://www.jev-llm.com/dashboard/). Keys are stored encrypted on the server, used only for runs you trigger, and never sent back to the browser. Need a Flatkey key? Get one for Jev 1.13 on Flatkey (https://flatkey.ai/models/typesafe~2Fjev-1.13).

Q: Is this the official TypeSafe Jev product page?

No. jev-llm.com is an independent research and practice site. Verify pricing, models and terms at the original TypeSafe sources.

Try it

Copy the code ↓

Data source
Decision schema
Model inference
Structured result

1 Data source Need, complaints, recent launches · editable

2 Decision schema 3 typed questions

need_typechoiceintegrationreliabilitypricingworkflow_gapusabilityother
severityscoreminorannoyingpainfulblocking
solved_by_launchesnoulprobability 0–1

3 Model inference typesafe/jev-1.13

Waiting for a run.

4 Structured result

Run the demo to see the JSON answers, the highlighted pick and the final action.

Call Jev yourselfSame data and schema, straight to Flatkey. Set FLATKEY_API_KEY and run.

curl -sS https://router.flatkey.ai/api/alpha/decisions \
  -H "Authorization: Bearer $FLATKEY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "typesafe/jev-1.13",
  "state": "Need: Stripe → QuickBooks invoice export keeps breaking\n\nSample complaints (38 in 90 days):\n- \"Every month-end the sync drops half the invoices and I re-enter them by hand.\"\n- \"Our bookkeeper spends a full day fixing duplicated Stripe payouts in QBO.\"\n\nRecent launches (12 months): 1 — a Zapier template for Stripe → Google Sheets.\n\nTask: need_type, severity, solved_by_launches.",
  "questions": {
    "need_type": {
      "type": "choice",
      "instructions": "What kind of product need do these complaints describe?",
      "criteria": {
        "integration": "Two tools do not connect, sync, import, or export data reliably",
        "reliability": "A product the user already has is buggy, slow, or loses data",
        "pricing": "The product works but costs too much or the pricing model does not fit",
        "workflow_gap": "No tool covers a step of the user'\''s job, so they do it by hand or in spreadsheets",
        "usability": "The capability exists but is confusing, hidden, or hard to set up",
        "other": "None of these"
      }
    },
    "severity": {
      "type": "score",
      "instructions": "How severe is the pain for the people complaining, judged only from what they wrote?",
      "criteria": [
        "minor",
        "annoying",
        "painful",
        "blocking"
      ]
    },
    "solved_by_launches": {
      "type": "noul",
      "instructions": "At least one of the listed recent launches already solves this need as the complainers describe it—true only when a launch clearly addresses the same problem for the same users."
    }
  }
}'

// Node.js 18+ (built-in fetch). Run: FLATKEY_API_KEY=... node jev.mjs
const body = {
  "model": "typesafe/jev-1.13",
  "state": "Need: Stripe → QuickBooks invoice export keeps breaking\n\nSample complaints (38 in 90 days):\n- \"Every month-end the sync drops half the invoices and I re-enter them by hand.\"\n- \"Our bookkeeper spends a full day fixing duplicated Stripe payouts in QBO.\"\n\nRecent launches (12 months): 1 — a Zapier template for Stripe → Google Sheets.\n\nTask: need_type, severity, solved_by_launches.",
  "questions": {
    "need_type": {
      "type": "choice",
      "instructions": "What kind of product need do these complaints describe?",
      "criteria": {
        "integration": "Two tools do not connect, sync, import, or export data reliably",
        "reliability": "A product the user already has is buggy, slow, or loses data",
        "pricing": "The product works but costs too much or the pricing model does not fit",
        "workflow_gap": "No tool covers a step of the user's job, so they do it by hand or in spreadsheets",
        "usability": "The capability exists but is confusing, hidden, or hard to set up",
        "other": "None of these"
      }
    },
    "severity": {
      "type": "score",
      "instructions": "How severe is the pain for the people complaining, judged only from what they wrote?",
      "criteria": [
        "minor",
        "annoying",
        "painful",
        "blocking"
      ]
    },
    "solved_by_launches": {
      "type": "noul",
      "instructions": "At least one of the listed recent launches already solves this need as the complainers describe it—true only when a launch clearly addresses the same problem for the same users."
    }
  }
};

const res = await fetch("https://router.flatkey.ai/api/alpha/decisions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.FLATKEY_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify(body),
});
if (!res.ok) throw new Error(`Jev request failed: ${res.status} ${await res.text()}`);
const { answers, usage } = await res.json();
console.log(JSON.stringify(answers, null, 2), usage);

# Python 3.8+ (pip install requests). Run: FLATKEY_API_KEY=... python jev.py
import json, os, requests

body = json.loads(r"""{
  "model": "typesafe/jev-1.13",
  "state": "Need: Stripe → QuickBooks invoice export keeps breaking\n\nSample complaints (38 in 90 days):\n- \"Every month-end the sync drops half the invoices and I re-enter them by hand.\"\n- \"Our bookkeeper spends a full day fixing duplicated Stripe payouts in QBO.\"\n\nRecent launches (12 months): 1 — a Zapier template for Stripe → Google Sheets.\n\nTask: need_type, severity, solved_by_launches.",
  "questions": {
    "need_type": {
      "type": "choice",
      "instructions": "What kind of product need do these complaints describe?",
      "criteria": {
        "integration": "Two tools do not connect, sync, import, or export data reliably",
        "reliability": "A product the user already has is buggy, slow, or loses data",
        "pricing": "The product works but costs too much or the pricing model does not fit",
        "workflow_gap": "No tool covers a step of the user's job, so they do it by hand or in spreadsheets",
        "usability": "The capability exists but is confusing, hidden, or hard to set up",
        "other": "None of these"
      }
    },
    "severity": {
      "type": "score",
      "instructions": "How severe is the pain for the people complaining, judged only from what they wrote?",
      "criteria": [
        "minor",
        "annoying",
        "painful",
        "blocking"
      ]
    },
    "solved_by_launches": {
      "type": "noul",
      "instructions": "At least one of the listed recent launches already solves this need as the complainers describe it—true only when a launch clearly addresses the same problem for the same users."
    }
  }
}""")

res = requests.post(
    "https://router.flatkey.ai/api/alpha/decisions",
    headers={"Authorization": f"Bearer {os.environ['FLATKEY_API_KEY']}"},
    json=body,
    timeout=60,
)
res.raise_for_status()
data = res.json()
print(json.dumps(data["answers"], indent=2), data.get("usage"))

No Flatkey key yet? Get one for Jev 1.13 →

About this use case

Use Jev to rank product needs: incoming state, Jev decision, routed outcome

Who / when

Audience

founders and PMs mining Reddit, reviews, and support threads for problems worth building
product-discovery and market-research teams scoring opportunity backlogs
indie hackers who want interview leads without paying frontier-model prices per row

Triggers

you have hundreds of complaint clusters and no consistent way to compare them
stakeholders want to re-weight growth vs. severity and see the ranking move
classifying every need with a chat model is projected to cost thousands of dollars
you need to know which pains already have recent launches and which are wide open

The workflow

01
Need + evidenceNeed summary, sample complaints, recent launchesInput
02
Need typeintegration / reliability / pricing / workflow …Jev
03
Severity + coverageHow painful, and do launches already solve it?Jev
04
Weighted rankCode mixes user weights with counts and growthYour code

Question shape

{
  "need_type": {
    "type": "choice",
    "instructions": "What kind of product need do these complaints describe?",
    "criteria": {
      "integration": "Two tools do not connect, sync, import, or export data reliably",
      "reliability": "A product the user already has is buggy, slow, or loses data",
      "pricing": "The product works but costs too much or the pricing model does not fit",
      "workflow_gap": "No tool covers a step of the user's job, so they do it by hand or in spreadsheets",
      "usability": "The capability exists but is confusing, hidden, or hard to set up",
      "other": "None of these"
    }
  },
  "severity": {
    "type": "score",
    "instructions": "How severe is the pain for the people complaining, judged only from what they wrote?",
    "criteria": [
      "minor",
      "annoying",
      "painful",
      "blocking"
    ]
  },
  "solved_by_launches": {
    "type": "noul",
    "instructions": "At least one of the listed recent launches already solves this need as the complainers describe it—true only when a launch clearly addresses the same problem for the same users."
  }
}

Gate policy

≥ 0.90

Write the labels; include the need in the weighted ranking and the interview-lead export.

0.65–0.90

Rank it, but mark the row for a quick human glance before it goes into interview outreach.

< 0.65

Keep it out of the ranking and send it to a review queue—usually the need summary mixes two problems.

Failure modes

Pre-launch checks before you trust auto-routing:

Letting Jev compute the ranking—counts, growth rates, and weights are arithmetic and belong in code;
Feeding a need summary that mixes two different problems, so every label comes back unsure;
Treating "a launch exists" as "solved" when the launch targets different users or a different step;
Asking a chat model to classify each row and paying hundreds of times more for the same labels;
Re-running classification every time someone moves a weight slider instead of caching the labels;

Keep exact work in code: complaint counts, growth over the study window, launch counts, the user's weights, and the final sort stay deterministic. Jev only answers the meaning-level questions per need—what kind of need, how painful, already solved?—so you classify once, cache the labels, and let people re-weight the ranking for free.

FAQ

Why use Jev instead of a chat model for this?

Each need is the same three short questions over a small block of text. That is a classification job, not generation. In the source post, classifying 149 needs cost about $22.51 in Jev credit versus an estimated $5.8k with existing LLMs—roughly 250× cheaper.

How do user-tunable weights fit in?

Jev labels each need once (type, severity, coverage). Your code turns those labels plus hard data—complaint count, growth, launch count—into numbers and computes a weighted score, e.g. w_growth × growth + w_severity × severity + w_gap × (1 − solved). Moving a weight re-sorts instantly with no new model calls.

What does the Flatkey call look like?

POST https://router.flatkey.ai/api/alpha/decisions with Authorization: Bearer …, Content-Type: application/json, and a body of state, model (typesafe/jev-1.13), and questions (choice / score / noul). This is not an OpenAI-compatible /v1 chat endpoint.

What should happen when Jev is unsure?

Low confidence usually means the need summary bundles two problems or the complaints are vague. Keep those rows out of the ranking, split or rewrite the summary, and re-ask—it is cheap enough to run twice.

Can I bring my own API key?

Yes. Try the free demo above without signing in. To run it live on your own text, sign in with Google and connect a Flatkey or OpenRouter key (a quick test call runs when you save it). Manage or delete it anytime under Dashboard → BYOK. Keys are stored encrypted on the server, used only for runs you trigger, and never sent back to the browser. Need a Flatkey key? Get one for Jev 1.13 on Flatkey.

Is this the official TypeSafe Jev product page?