Written Identity
The project describes a written identity and principles that sit beside a model. The page does not list tested model adapters or show how behavior compares across models.
An internal agent project with a written Lighthouse doctrine and a Pilot work-order concept.
Hermes is the name Lighthouse Agentic uses for its internal conductor-agent project. This page does not claim a connection to other products with a similar name. It describes the project's design; it does not list tested model adapters, provider data flows, or controls enforced in every setup.
Learn moreHermes is designed around a written identity and principles that sit beside a chosen model. This page does not compare measured behavior across model changes.
The written doctrine describes Hermes goals and principles. The project also describes an ai-brain, meaning a structured store of knowledge. This page does not show how that material is loaded in a run or which models are supported.
A Pilot Brief is a work order that states a job's goal, scope, and limits. This page describes Briefs as part of the Hermes and Pilot design. It does not show a signed Brief being checked in a live run.
The project describes a written identity and principles that sit beside a model. The page does not list tested model adapters or show how behavior compares across models.
The project describes an ai-brain as structured knowledge for Hermes. The page does not show its contents, update process, or how sources are checked during a real task.
The design places a Pilot Brief before work and describes checks around scope. The page reports Pilot Lite as shipping for internal work and Fortress as in development. It does not show that either tier enforces every control in a live run.
The design goal for Hermes is to become the exclusive runtime for Lighthouse agentic work: the single, governed conductor through which every job passes. That is where we are building toward. It is not where we are today.
The Lighthouse doctrine sets out a worldview and commitments for the Hermes project. This page describes the written document; it does not show whether or how it is loaded during real runs. The three pillars below reflect that stated doctrine.
The written doctrine calls for respect for each person's choices.
The written doctrine calls for honest limits. This page does not include an evaluation of answer accuracy or confidence.
AI is a tool; God is the source. The written doctrine places the work under that faith-led view.
This comparison summarizes the project's stated design. It does not compare tested model behavior or provider data handling.
Status recorded in the October 6, 2026 audit: the page reports internal use and a Pilot integration. The public page does not list tested model adapters or show what each provider receives.
The page describes a written document with Hermes's stated principles, worldview, and identity. It does not show how often that document is used in real runs.
The page reports Lite as shipping for internal work and Fortress as in development. Pilot's explainer does not include a test record for each control.
The design goal is for Hermes to become the singular runtime for all Lighthouse agentic work. We are building toward that. We are not there yet.
Hermes is described here as an internal project, not a standalone product offer. For a conversation about Lighthouse Agentic work, Start a Conversation.
A written identity is part of the Hermes design.
The project describes an internal conductor-agent design. Its current enforcement scope is not shown here.