Migrate from Copyleaks

Why switch

• Synchronous answer. 80ms warm latency vs Copyleaks' 30-90 second scan queue. Drop ~150 lines of webhook plumbing.
• Single API key. Bearer token vs Copyleaks' email + key + access-token-exchange dance.
• Per-passage breakdown. Plus tier ships paragraph + sentence highlights with the fired rule ids. Copyleaks shows match-only.
• Lower cost. $0.0003/word or $0.50/call vs Copyleaks' $0.15-$0.36 per page.
• Hidden multilingual. 20+ languages — no per-language quota or flag.
• Privacy default. Plaintext NOT retained.

Side-by-side

// Before: Copyleaks (async — submit + wait for webhook)
// 1. Get a token
const token = await fetch('https://id.copyleaks.com/v3/account/login/api', {
  method: 'POST',
  body: JSON.stringify({ email: process.env.CL_EMAIL, key: process.env.CL_KEY }),
});

// 2. Submit a scan with a callback URL
await fetch('https://api.copyleaks.com/v3/businesses/submit/file/SCAN_ID', {
  method: 'PUT',
  headers: { Authorization: 'Bearer ' + token.access_token },
  body: JSON.stringify({
    base64: Buffer.from(text).toString('base64'),
    filename: 'doc.txt',
    properties: {
      webhooks: { status: 'https://your.app/cl-webhook' },
      sandbox: false,
    },
  }),
});

// 3. Wait for webhook callback. Receive results in a separate handler.
// (50-200 lines of webhook plumbing to actually get the answer.)

// After: deepaidetector (synchronous — answer in the response)
import { createClient } from '@deepaidetector/client';
const client = createClient({ apiKey: process.env.DEEPAIDETECTOR_KEY });

const data = await client.detect({ text });
const aiScore = data.score;          // 0-1
const verdict = data.band;           // categorical
// You can still subscribe to webhooks (for fan-out to other systems),
// but the synchronous answer is right here.

Field mapping

Behaviour differences to know

• Authentication is one step. No token-exchange — bearer auth directly. Save your dad_live_ key as a secret, that's it.
• Synchronous. The answer is in the response body. You can still wire webhooks for fan-out, but they're not required for the primary flow.
• Base64 not needed. Send the text as a UTF-8 string in the JSON body.
• Detection + plagiarism are separate endpoints. Call both in parallel for the bundled equivalent.
• 80-word minimum. Below 80 words on /v1/detect we return 422 insufficient_text.
• Different units for plagiarism overall. Copyleaks: 0-100 integer. Us: 0-100 float (overallPercent).

Migration checklist

1. Generate a Deep AI Detector API key in your dashboard.
2. Delete the Copyleaks token-exchange step entirely.
3. Delete the Copyleaks webhook handler (or leave it for fan-out — your call).
4. Replace the submit + wait flow with a synchronous client.detect() call.
5. Replace base64 + filename body with plain JSON containing text.
6. Update response parsing per the field map.
7. Add 422 handling for short text.
8. If you still need plagiarism, add a parallel client.plagiarism() call.