System:
A hardware system you’re building, operating, or analyzing in Atlas. A system spans both its design (schematics, layouts, BOMs, software, specs) and the assets built from it (the physical units running in the field). Designs define how the system should behave; assets reveal how it actually behaves.Metagraph:
Atlas’s living model of a system. Nodes represent the elements that make up the hardware (subsystems, components, interfaces) along with the design context and operational data attached to them. Edges describe how those elements relate. The metagraph is what grounds Atlas’s reasoning in your actual hardware rather than guesswork.
Session:
An interactive conversation with Atlas. Bring questions, share context, and iterate on analysis. Sessions are where exploratory and open-ended work happens.Job Run:
A single execution of an asynchronous unit of AI work. Job runs are how longer, structured tasks get done in the background, and they produce outputs like issues, analysis reports, and updates to the metagraph.
Issue:
A problem Atlas has identified in a system: a failure, anomaly, root cause, or open action item observed on an asset or surfaced across the fleet. Issues carry status, priority, comments, and links back to the data they came from.Test Run:
A record of a verification campaign against a design or asset. A test run gathers individual test cases, the data behind them, and Atlas’s analysis of how things performed.Attachment:
A file surfaced by Atlas during a session or job run. Attachments can be annotated and exported through a signed download URL.
Memory:
Knowledge Atlas carries across interactions. System memories are attached to a system and capture the things that aren’t written down anywhere: behavioral quirks, design decisions, known failure modes. Personal memories are attached to a user and capture how that individual likes to work.Skill:
Encoded domain expertise, scoped to a user or user group. Skills are reusable instructions Atlas draws on while reasoning, capturing how you or your team approach specific kinds of problems.
