| What it does | On-device moderation that flags hateful, abusive and threatening text in 23 European languages, before the message is sent.
| Independent lab testing how AI agents access real software
|
|---|
| Stage | MVP
| Startup
|
|---|
| Founded | Not given
| 2026
|
|---|
| Based in | NL
| NL
|
|---|
| Pricing model | Freemium
| Freemium
|
|---|
| Pricing | Free up to 100k monthly active devices with unlimited inference. Custom licenses available.
| Free check (one per domain every 90 days). Watch: $49/month or $490/year (up to 3 products, includes private monthly drives and fixes). Builder: $49/month (follow any number of tools with level change notifications). Team: $290/month (includes Builder plus listing and history file). Tool comparison:
|
|---|
| Key features | - Three separate scores: hateful, abusive, threat
- Ten target signals under the hate score: race, color, religion, descent, national or ethnic origin, sexual orientation, gender, disability, age, other
- Runs fully on device with no server calls
- Supports 23 European languages from one model file
- 80MB model size optimized for iOS, Android, and browser
- 0.847 macro F1 accuracy on Multilingual HateCheck
- Warn before message is sent
- Comment triage without infrastructure
| - Tests AI agents across 9 checkpoints from documentation to account deletion
- Publishes directory of 245 products with access levels (L0-L5)
- Free check shows what agents encounter on your product
- Watch subscription provides monthly private drives with findings
- Open data available as JSON, Markdown, and over MCP
- API and MCP server for accessing drive data
- Monthly reports tracking agent access trends
- Identifies where agents stop (sign-up, verification, bot checks, payment)
- Documents specific findings with evidence and fixes for owners
|
|---|
| What makes it different | - Multilingual support across 23 European languages in a single 80MB model
- Complete privacy with no data leaving the device
- Free access up to 100k monthly active devices with unlimited inference
| - Only independent lab publishing public record of AI agent access across real software
- Tests with real agents automatically driving complete workflows, not just APIs
- Covers 245 products with consistent methodology and 9-checkpoint testing
- Levels confirmed by repeated drives on different days for accuracy
- Payment transparency: owners pay only for private drives, not for public ratings
|
| Community votes | 0 | 0 |
| DR (Domain Rating by Ahrefs) | 38 ▲ +1 (up) since last check | 2 ● no change since last check |
| Trust Flow (Majestic) | 20 | 4 |
| Citation Flow (Majestic) | 39 | 25 |
| Referring domains (Majestic) | 616 | 29 |