ABOUT

Solace Systems

We’re a small AI lab. We research how agents fail, build the harness that stops them, and ship it: as Solace AI, as a platform for developers, and as services for businesses.

Concept illustration of a device-testing workbench with Android phones and a notebook
RESEARCH ON REAL DEVICESConcept artwork · Not a photograph of our lab

The harness is the work.

An agent is only as good as what surrounds the model. The same model can be careless or careful depending on the harness: what it can touch, what it has to prove, and what it isn’t allowed to do. The model will keep changing. The harness is the work.

How we work

  • We test on a real phone. The harness is developed and tested on Android hardware.
  • Every rule has a failure behind it. When Solace gets something wrong, we fix the class of mistake and write down why.
  • We publish what we measure. We say which numbers we don’t have yet.
  • The model is swappable. Solace isn’t tied to one AI provider.

Get in touch

General: hello@solace-systems.com
Press: press@solace-systems.com
Security: security@solace-systems.com

Founder biography and registered company details are awaiting confirmation.