In Parallel keeps a rich, shared understanding of each Workspace current for people and AI. Intended outcomes can be part of that context within Workspace pages rather than a separate tool-like Goals surface.

A Workspace is the unit of accountability and context in In Parallel. It sets the boundary around a team, project, program, operation, or account: who can see it, what work belongs together, and which context AI may use.

A Workspace is usually right-sized when one person is accountable for its outcomes, its decisions happen in a coherent cadence, and its priorities can be ranked together. If updates feel noisy or contradictory, split the Workspace by accountability or cadence.

Shared context is the maintained understanding of a Workspace that people and AI can use without reconstructing it from scattered meetings, documents, and messages.

It has several layers:

These layers are connected, but they do not all have the same role. Authorized sources provide evidence. Signals identify meaningful change, and observations retain structured, source-linked details. Maintained pages turn those signals and other Workspace material—including goals or other intended outcomes—into a useful narrative. Topics, briefings, and any plans a Workspace uses help people act on the context.

Workspace pages are the durable, shared context layer inside a Workspace. Their purpose is practical: a new team member, returning colleague, or authorized AI should be able to understand the current situation without reading every meeting transcript.

In Parallel can maintain pages such as a Charter, Objectives page, Runbook, Retro Log, or meeting page. People can edit those pages directly. About this page explains a page's purpose, perspective, source inputs, and refresh behavior.

A maintained page is not a claim that AI-generated prose is automatically true. Check the update or refresh state and follow source records before relying on a consequential claim.

Goals remain important as context: they state what the Workspace is trying to achieve and how progress will be judged. They can be described and maintained within Workspace pages; there is no longer a separate tool-like Goals footer for managing them.

A goal does not need to carry the full explanation of the Workspace. Maintained pages can hold the intended outcomes together with the mission, scope, assumptions, decision rights, operating model, and relevant history. This gives AI richer context for interpreting whether work supports the intended outcome.

When the Recorder joins a meeting, it captures Notes and detects signals such as decisions, actions, risks, questions, learnings, and ownership changes. Connected documents and authorized Slack or Microsoft Teams conversations can provide additional source context.

The transcript or conversation remains source material. Structured observations retain their source relationships and contribute to the Workspace's shared context. This gives people and AI evidence for detecting change, updating maintained context, and checking a claim.

Signals are meaningful information from meetings and other authorized sources. A Workspace's observation kinds define what AI should look for, such as decisions or a customer-defined kind like customer need. Findings refers only to the product section where observations may still be listed; that section is being phased out.

Each meeting deepens the shared context. A new decision can be interpreted alongside the Workspace's existing charter, goals, previous decisions, and open risks rather than as an isolated note.

AI uses the authorized slice of shared context that fits the task:

The result is not one giant prompt. Context is assembled for the reader and task, subject to the same Workspace permissions as the underlying records.

AI may maintain narrative context automatically, but that is different from silently making a decision. People remain accountable for decisions, ownership, communications, and other consequential actions.

This is an expansion, not a claim that goals or plans no longer matter. They become more useful because they sit inside context that also explains why the work exists, how the team operates, what has changed, and which constraints matter.

The current mental model:

When the loop works, people and AI start from the same maintained context. Less time goes into reconstructing what the work is, and more time can go into judgment and action.