Let your team use AI.
Without leaking your customers.
Aegisly sits between your apps and the LLM. It strips emails, card numbers, national IDs and API keys before they leave your company, enforces per-team policies and budgets, and records every call in a tamper-evident audit log your compliance team can verify.
One endpoint change. Full control.
Point your app at Aegisly instead of calling the model directly. Every request runs through the same policy pipeline.
Redaction before the model
12 detectors: email, phone, Luhn-validated cards, US SSN, Brazilian CPF, IBAN, IPs, AWS/GitHub/Stripe-style keys, JWTs and private keys. Sensitive values become placeholders like [EMAIL_1].
Policies per workspace
Choose redact, block or off for PII and secrets, ban confidential terms ("Project Falcon"), cap prompt size and pick allowed models — changes apply instantly.
Tamper-evident audit
Each entry stores a SHA-256 of the prompt (never the prompt itself) and the hash of the previous entry. One click verifies the whole chain; CSV export for auditors.
Keys & budgets per team
Issue a key per app or department with its own monthly budget. Revoke instantly. Keys are stored only as hashes.
Usage & cost visibility
See requests, blocks, redactions, tokens and estimated spend per key for the month — the numbers finance and security both ask for.
Built on the edge
Cloudflare Workers + D1 + Workers AI. Low latency worldwide, no servers to patch, open-source codebase.
Pricing
Start free. Upgrade when AI becomes part of how your team works.
Create your workspace
Free plan, no card. You get an admin key for the dashboard and a member key for your first app.