Mental Models
Reasoning tools for understanding why AI engineering needs shared, evidence-backed software system models.
Mental models explain why Workspace Intelligence exists before they explain how any implementation works. They help teams distinguish files from systems, context from understanding, agent memory from shared truth, and workflow automation from an intelligence layer.
The central model
The model is shared; each consumer receives a view appropriate to its task. Consumers may propose changes, but they do not independently redefine system truth.
Read the models
- Shared System Model — one governed model, many consumer views.
- Why AI Models Should Not Own the System Model — generation and authority are different responsibilities.
- Workflow vs Workspace Intelligence — execution graphs do not replace system understanding.
- Evidence-backed Context — context becomes trustworthy when facts retain evidence, scope, freshness, and confidence.
- Why Every Agent Needs the Same Workspace — agents need a common substrate without receiving identical prompts.
- From Repository to Workspace Model — the transformation from source containers to a software-system view.
How to use this section
Use these pages to evaluate architecture claims. If a tool says it understands a system, ask:
- What is the canonical model?
- Which evidence supports each fact?
- Who can mutate the model?
- How are conflicting observations resolved?
- Can every consumer identify freshness, scope, and uncertainty?
Those questions are independent of Workspai. They define the category boundary.