E.M.M.A. governance
Every action Lucy proposes goes through E.M.M.A.: what Lucy may do is defined by capability grants, owner approvals and consent records, and E.M.M.A. keeps a tamper-evident record of what happened.
◐ Prototype / under testing
Fenton Lab builds a governed AI operating system, Lucy AIOS, and independent software around it. This page shows the real interfaces where they exist and says plainly what each piece does today.
Fenton Lab is sharing authentic development previews of the Lucy ecosystem. These applications are actively being engineered and tested. Screenshots may show demonstration data, incomplete features, or capabilities awaiting validation. Displayed functionality does not imply commercial availability.
Nothing here is available to buy, download or use yet. Everything shown is in development on Fenton Lab's own machines. Screenshots come from the running software with fictional or freshly created demonstration data.
Lucy decides what work matters. E.M.M.A. decides whether it may run. R1 decides what resource posture the machine can safely support.
LUCY AIOS · GOVERNED CORE
CONNECTED BY A SIGNED BRIDGE · IN DEVELOPMENT
Lucy ProjectsClaims & Evidence ManagementINDEPENDENT · NO LUCY GOVERNANCE INSIDE
Glimwynn · EyravonDesktop companions with their own identity and permissionsA Windows-native, local-first AI operating system designed around human oversight, recorded evidence and clearly defined execution boundaries.
◐ Prototype / under testing Not available to buy, download or use.
Every action Lucy proposes goes through E.M.M.A.: what Lucy may do is defined by capability grants, owner approvals and consent records, and E.M.M.A. keeps a tamper-evident record of what happened.
◐ Prototype / under testing
R1 shapes work to what the machine can safely support, alongside Lucy's memory-pressure policy, and shows CPU, memory and disk readings in the dashboard.
◐ Prototype / under testing
Read-only security observation: it watches the host and incoming instructions, and records security facts as evidence. Eagle Eye observes only; it never approves or runs work. Today it checks engineering work arriving from Helix.
◐ Development-stage back-end component
An engineering-work pipeline. Helix can only propose bounded work; Lucy owns the task identity and the build ticket, the owner approves through Lucy's normal approval step, and Lucy's own executor runs it.
◐ Development-stage back-end component
Eagle Eye and Helix have no user interface of their own, so they are described rather than pictured.
An experimental application for organizing projects, documentation and evidence workflows, with a tamper-evident audit trail and human sign-off where it matters.
◐ Prototype / under testing Shown with a fictional demonstration project only.
A browser-based coding workspace with a code editor, AI companion panels and a registry of engineering agents. It runs on Fenton Lab's own machine.
◐ Prototype / under testing Shown with a fictional sample project only. No AI service was connected or called for these screenshots.
Companions are separate products with their own identity and permissions. They don't contain Lucy's governance components.
An expressive fox companion with local memory, privacy controls and optional AI providers, being prepared for a controlled, invitation-only private beta.
◐ Prototype · private beta preparation
A watchful eagle companion with a calm, deliberate personality and its own independent identity.
◐ Prototype · in development
Pictures of the companions will be added once their artwork is approved for the web. See the companion collection.
Work that is under way or designed, but not yet running as release software.