How We Provision Secure Hermes Agents
How we provision and run a fleet of Hermes agents on hardware we own: one agent per job, each locked to its own memory and tools, reachable over Signal, with your data staying inside your environment.
Articles
Technical case studies from the field.
Field notes and engineering case studies from real client engagements. What we built, why we built it that way, and what has held up over time.
Running Enterprise AI On-Prem: Our Self-Hosted Inference Stack
How we built a private inference cluster that runs state-of-the-art open-weight models on hardware we own, serves a unified OpenAI-compatible endpoint, and keeps every prompt behind our own firewall. Zero data egress. Sub-second latency. 262K token context windows.
Building a Fault-Tolerant, Multi-Tenant Platform on AWS
How we engineered a self-healing cloud infrastructure for Educational Partners International that has been running without incident since July 2020. No failover scripts. No 3 a.m. pages. Just infrastructure that absorbs failures and keeps going.