Why Hermes Labs exists.
Hermes Labs does AI reliability engineering. We find where AI agents and the systems they run on fail before production. Not every agent failure is an AI failure. Sometimes it’s a system failure. We investigate model behavior, instructions, memory, tools and runtime code, then engineer the fix the evidence supports. We call the research practice Epistemic Engineering.
The founder
Hermes Labs was founded by Rolando Bosch, who goes by Roli. He studied philosophy in college, with a lasting focus on philosophy of language. His technical path has been self-directed, bringing those questions about meaning and interpretation into hands-on work with AI systems.
Before turning to this work, Roli spent about three years in the nonprofit sector. That background, together with his study of how meaning depends on context and interpretation, shaped the questions he brings to AI systems: what survives when language is summarized, retrieved, or handed from one agent to another, and what evidence remains when it does not?
Hermes Labs grew from that combination of philosophy and hands-on engineering. The lab focuses on the places where a system can remain fluent while losing the instruction, evidence, or context it was supposed to preserve.
The name points to hermeneutics, the study of interpretation and understanding, and to Hermes, the messenger and translator. It is a direct description of the work: making the language layer of AI systems more inspectable and accountable.