Skip to content

Tabs and workspaces

This page explains the visible top-level product surfaces for end users. The Settings tab is intentionally left out. Use it when you want to understand what each main tab is for before you start clicking through the product. It is most useful for new users who can already sign in but do not yet know where to go next. A quick read usually takes three to five minutes.

Note

Some tabs are visible only for certain roles or feature configurations.

How to use this guide

Each guide page answers a different kind of question:

Top-level tabs

Projects

Use Projects to create a project, choose the project language, and set its access mode. From a project page you get the high-level system map:

  • the Specification (What) side shows specifications, Epics, Use Cases, and Requirements
  • when enabled, the Planning (Ready) card shows spec-scoped Plans, Scopes, Estimates, Risks, and Effort drivers
  • the Implementation (How) side shows implementations, features, architecture, builds, delivery packages, Targets, and instruction sets

The project detail page is also where you open project-scoped lists of specifications, implementations, and Plans. In practice, this is often the first tab people open when they are orienting themselves in a new tenant or checking which project they should work in.

Specifications

Use Specifications to browse specification items across the accessible project scope. For most analyst work, open Workspace on a specification row first; it gives a guided preparation path before sending you into detailed Epics, Requirements, Use Cases, Test Cases, Ubiquitous Language, and Gap analysis tools. In the cross-project view, specifications are grouped by project and show their key and description to make selection easier. Search filters the accessible result set before paging, and Load more continues through additional matches.

Implementations

Use Implementations to browse or create implementation containers. From an implementation you manage features and, when enabled, build jobs. Builder-side work normally starts here unless you already entered through a project page. In the cross-project view, implementations are grouped by project and show their key, linked specification, and description when available. Search filters the accessible result set before paging, and Load more continues through additional matches.

Plans

Use Plans to browse planning containers. New Plans are scoped to a Specification and contain one Planning Scope for each selected Epic. Each Scope owns the AI-backed Assessments for that planned part and has its own detail page for generation, review, overrides, and baseline acceptance. Plans are available only when Planning is enabled and your role allows it. Assessments run in the background and produce separate Full and Active Scope analyses plus a canonical per-item effort split. Accepted Scope estimates and immutable Minimum Baseline revisions expose freshness and explicit reconciliation. They record the estimation profile version used and remain auditable when failed or superseded. Follow the Planning flow when you need the step-by-step path from specification scope to an accepted planning baseline and optional effort allocation. Plans require CAP_PLANNING and are visible to Admin and Planner roles.

AI Usage

Use AI Usage when you need a quick operational view of how AI is being used across the tenant. This is the place to answer practical questions such as: "Did usage spike today?", "Which feature or model caused it?", or "Have we already exhausted the included monthly budget?" You do not need any setup beyond access to the page, and a quick review usually takes one or two minutes.

From this tab you can:

  • see estimated cost over time
  • check input and output tokens together with estimated cost
  • compare usage by feature and by model
  • review tenant AI billing for the selected UTC billing month
  • move between UTC billing months with previous, next, and current month controls
  • see how included budget usage is tracking against the monthly budget
  • check the remaining included budget or the additional amount charged for premium requests in the selected month
  • review premium request count, premium charge, and billable spend
  • download the monthly usage report with project key and project name columns
  • see whether premium brokerage is already active or will only apply after the included budget is exhausted
  • review recent AI events

This tab is for operational visibility, not artifact editing.

About

Use About to confirm the deployed version, edition, and active feature flags. This is useful when behavior differs between environments and you need to confirm whether you are looking at the same capabilities as another team or deployment.

Tenant

Use Tenant to see the active tenant context. For authorized users it also shows:

  • tenant users
  • current tenant roles, active or inactive status, and last active time for each listed user, with a user-name filter and Load more control when more than 20 users match
  • for Manager users, a create-user flow with validation for username, email, first name, last name, temporary password, and assigned roles
  • for Manager users, a deactivate action for active tenant users
  • tenant format presets that define output standards for future AI generations; duplicate a preset to create another one and keep preset names within 16 characters; delete persists after confirmation unless a project still uses that preset

This tab is mainly administrative, so most end users visit it only when they need to manage users, format presets, or shared tenant-level configuration.

AI prompt selectors use the active tenant’s settings. Without an explicit override, each action uses its own default prompt (for example, Use Case generation and enhancement have separate defaults). Enhancement applies the project’s selected format preset, or the tenant default when no project preset is selected.

Profile

Use Profile to review your user details, language preference, current claims derived from your access token, and saved AI quick phrases. Saved phrases are grouped under Architecture workspace phrases by action and Architecture workspace tab, and under Specification workspace phrases by Requirements, Use cases, and Test cases. This is also where you mark phrases as favorites or delete them. It is a short page to review, but it becomes useful over time if you repeat the same AI prompts and want to reuse them instead of rewriting them.

Contextual Assistant evidence

The Specification and Implementation Assistants resolve your question, follow-ups and corrections before selecting evidence. They retrieve only the planned artifact families and fetch detailed content when needed. Published Organization evidence is searched only when the question calls for it. Retrieved excerpts are assessed for relevance before the answer; citations refer to evidence supplied for that turn.

Collection counts are complete only when pagination establishes the full collection. A partial result is not a total, and shortened artifact details do not establish complete content coverage. Answers state whether evidence is focused, complete collection metadata, partial or insufficient. An ambiguous request can ask for clarification before searching.

Implementation first validates the focused artifact and its ancestors. It can then follow selected links, such as Feature → Use Case → Requirement or a current Architecture reference. Traversal is bounded to two hops and a shared retrieval budget. Unavailable references, unresolved human references and exhausted limits are reported as incomplete chains. Retrieved links do not establish factual support by themselves. Existing proposal review and approval controls still govern changes.

AI Assistant prompts

Project Workspace, Project setup, contextual Assistants (including Support Workspace), and Support Case show a thin status strip beneath the submitted prompt, with smaller text and animated dots throughout processing. The animation remains during conversation refreshes and stops when the operation completes or fails. Project Workspace includes the current stage and coverage counts in this strip. Use the ▤ glyph to open results, including from a completed response. Glyph controls have hover tooltips and screen-reader labels. Failed turns retain their error and the ↻ retry control. The strip never exceeds its prompt’s width. All status and error text is right-aligned, wraps within that width, and clips any horizontal overflow.

The detected Intent appears beneath each prompt as soon as its validated planning result is available, while the animated status continues through retrieval and answer generation. Project setup derives intent in the same model response as its answer and displays both together. Intent is shown only beneath the user prompt; the UI hides redundant derived-intent headings and their metadata JSON in both existing and new Assistant responses, including copied responses.

Assistants interpret the latest request using recent complete conversation turns and the current draft or proposal state. In a supported review workflow, “create”, “save” or “update” can prepare a change for review; applying or publishing it still requires the existing approval. If the target remains ambiguous, the Assistant asks one focused question before searching or changing anything. For an unavailable action, it returns the specific reason only. Project Setup keeps its draft unchanged when clarification is needed or the request is unsupported; the Support Workspace Assistant remains read-only.

In contextual Assistant responses, click a citation or its Retrieved evidence card to see the source text captured for that response. Organization evidence shows the source title and the original excerpt. Earlier citations keep their own captured text even when a later response uses the same citation label.

Submitted Assistant messages are remembered in your personal prompt library within the current tenant. Open Profile → AI Assistant prompts to favourite or delete them. The four groups are Project Workspace, Project setup, Contextual, and Support Case. Project Workspace and Project setup have separate histories and favourites. The Contextual tab groups prompts by their launch context, such as Specification, Requirements, Epics, Implementation, Feature, Architecture or Support. Each contextual Assistant picker shows only its own context’s saved prompts. Related views of the same context share a group (for example, Implementation and Implementation Workspace). Previously saved contextual prompts without launch metadata remain manageable under Unclassified and are not mixed into a context-specific picker. Each group (and each launch context within Contextual) retains 20 recent non-favourites and all favourites; unfavouriting an older prompt can remove it from recent history.

An empty conversation shows up to five saved prompts, favourites first, and separate Examples. Click the small ▴ glyph above the composer or press Arrow Up while editing a message to open the searchable prompt picker. The glyph has a persistent tab outline joined to the composer divider and points down (▾) while the picker is open. Click it again to close the picker. Arrow Down or Escape closes it and returns focus to the message; selecting a prompt also closes it. Selecting a prompt inserts it at the last cursor position or replaces selected text without sending. Manage prompts, inside the picker, opens the corresponding Profile group in another tab. Edit a prompt in the composer and submit it to remember the revised text.

Only your new, successfully submitted messages are recorded. Conversation history, other users' messages, responses and retrieved evidence are not imported. Reuse keeps the current Assistant's context and permissions. Prompt saving errors do not resend or cancel your message. With authentication disabled, the library uses this browser's local storage; authenticated sessions use your Profile and never switch to local storage on errors.