Early access

Twelve teams in 2026. Verifiable evidence in a week.

Aegis is in private research preview. Tell us about your stack and your threat surface — we’ll come back with a one-page proposal: which Aegis layer to start with, what we’ll instrument, and what evidence you’ll have to show after week one.

Apply

Tell us what your agent touches.

Submitting opens an email draft addressed to the Aegis team — no servers, no analytics, no wait. Edit it before you send.

What partners get

First-cohort tooling
Sentinel binary, Cleanse policy starter kit, and a sealed SecureBench run for your stack.
A hardening review
We sit with your team and produce a written risk audit of your current agent stack — your stack, your threat model.
Direct line to the builders
Shared Slack with the Aegis team. Bug fixes ship the same day. Roadmap moves on what you actually need.
Co-authored case study
When the work’s done, you decide whether we publish — and on what terms. Public case study, private playbook, or both.

WHO QUALIFIES

01You ship an agent that interacts with the open web, third-party tools, or persistent memory.
02You can dedicate ~2h/week of an engineer or security lead for 4 weeks.
03You can run a single binary or attach a process supervisor to your agent runtime.
04You’re willing to share a redacted incident or threat model that helped us shape SecureBench.

RESPONSE WINDOW

We respond within 3 business days. Cohorts close on the 1st and 15th of each month.