Security & Trust
Everything your IT team needs to know
Overview
Loamly offers two deployment modes: a managed proxy for 100% AI traffic verification, and a JavaScript tracker for website analytics. This page covers data handling for both.
TL;DR: We never store IP addresses, request bodies, or response content. The managed proxy only logs AI bot visits. The JavaScript tracker logs all pageviews for analytics.
How it works
When you point your domain's DNS to Loamly:
User Request → Your Domain (A: 37.16.7.18)
→ Loamly Proxy (Frankfurt, EU)
→ SSL Termination (Let's Encrypt)
→ RFC 9421 Signature Verification
→ Forward to Your Origin (unchanged)
→ Response back to UserWhat happens at each step
| Step | What happens | Data accessed |
|---|---|---|
| SSL Termination | Caddy provisions and renews Let's Encrypt certificates automatically | Domain name only |
| Signature Check | We inspect HTTP headers for RFC 9421 signatures from AI bots (ChatGPT, Perplexity, etc.) | Headers only |
| Verification | Cryptographic verification using embedded public keys (JWKS) | Signature headers |
| Forward | Request forwarded to your origin server unchanged | Pass-through |
| Logging | If AI bot detected, we log: URL, timestamp, bot type, verification result | Metadata only |
Pass-through architecture
Data handling
What we collect
Managed Proxy (Edge Worker)
AI bot requests only. Human traffic passes through unlogged.
- Landing page URL
- Timestamp
- AI bot type (ChatGPT, Perplexity, Claude, Gemini)
- Verification result (signature valid/invalid)
- Country (derived from IP, IP discarded)
- User-Agent header
JavaScript Tracker (t.js)
All pageviews — both human and AI traffic, for analytics.
- Page URL, referrer, UTM parameters
- Timestamp, session duration
- Country (derived from IP, IP discarded)
- Device type, browser (from User-Agent)
- AI source classification (ChatGPT, Claude, etc.)
Data retention
- Free tier: 90 days
- Grow tier: Forever (no deletion — your data stays as long as you're a customer)
What we don't collect
- No IP addresses stored — Hashed for deduplication, then discarded
- No request bodies — Form submissions, API payloads pass through unread
- No response bodies — HTML, JSON, images not cached or logged
- No cookies — Uses sessionStorage, not cookies
- No PII — No names, emails, or personal data collected
Privacy by design
Open source
Loamly is fully open source under the MIT license:
- Main repository: github.com/loamly/loamly
- Edge verification: packages/edge — Cloudflare Worker for RFC 9421 signature verification
- JavaScript tracker: packages/tracker — Website analytics with AI detection
RFC 9421 verification uses Ed25519 cryptographic signatures. We verify against OpenAI's public keys (JWKS). Zero false positives.
Instant rollback
Removing Loamly takes 30 seconds. Just change your DNS record back to your original server:
# To remove Loamly:
# Change your A record from:
example.com A 37.16.7.18
# Back to your origin:
example.com A [your-original-ip]
# Or CNAME to your host:
example.com CNAME your-site.vercel.appDNS propagation typically takes 5-30 minutes. There's no lock-in, no migration process, no data to export.
Compliance
| Standard | Status |
|---|---|
| GDPR | ✅ Compliant — EU data residency (Frankfurt), no PII stored |
| CCPA | ✅ Compliant — No personal information sold or shared |
| SOC 2 | Not yet — Happy to answer security questionnaires |
| HIPAA | N/A — We don't handle health data |
Infrastructure
- Proxy: Fly.io (Frankfurt, Germany)
- Database: Supabase (Zurich, Switzerland)
- SSL: Let's Encrypt (auto-renewed)
FAQ
Can you see my customer data?
No. Request and response bodies pass through without being read or stored. We only inspect HTTP headers for AI bot signatures.
Can you modify my website content?
No. We're a transparent proxy. Your HTML, CSS, JavaScript, and API responses are forwarded byte-for-byte unchanged.
What if Loamly goes down?
Our proxy runs on Fly.io with automatic failover. In the unlikely event of an outage, you can point your DNS back to your origin within minutes. We target 99.9% uptime.
Do you cache my content?
No. We don't cache anything. Every request goes directly to your origin. Your origin's caching headers are passed through unchanged.
Can I self-host the proxy?
Yes. Contact us for the self-hosted deployment guide if you require on-premise infrastructure. The verification logic is already open source.
How do I verify you're not logging everything?
Our verification code is open source. For enterprise customers, we offer audit logs and can provide infrastructure access for security reviews.
Questions?