Brief a campaign lead and connect audience research, strategy, copy, design, and approval on one canvas.
Outcome: A campaign with visible owners, deliverables, and reviews from concept to publication.
Organize creative direction, references, images, accessibility, and consistency review as a specialist team.
Outcome: Documented visual decisions and artifacts connected to their creators and reviewers.
Coordinate research, briefs, writing, editing, and distribution through custom Kanban stages.
Outcome: A complete editorial pipeline with preserved context and human approval.
Use any provider you already have installed; Orkestrai detects the CLI, preserves its conversation, and presents every agent in the same visual workflow.
Outcome: Vibe coders, designers, marketers, and product teams work by outcome instead of command.
Choose any supported service from one compact Agents menu and pin up to four ready favorites in your preferred order.
Outcome: The agents you use most stay one click away across every workspace and app restart without hiding the rest.
Keep Claude, Codex, and Kimi usage on the canvas and configure a source, fallback, 5-hour/weekly/monthly window, and threshold shared with the leader.
Outcome: New work goes to a healthy provider without silently moving tasks that are already running.
Choose three dark themes or one light theme, then duplicate and edit semantic tokens with immediate preview.
Outcome: A consistent personal theme that can be exported and reused on another installation.
Switch the provider from the agent header while keeping its name, role, Maestro status, floor, position, and connections.
Outcome: Move a responsibility from Claude to Codex, Kimi, or another installed CLI without recreating the team.
Brief one leader. It proposes the team, recruits specialists, connects context, and distributes the backlog.
Outcome: A feature moves from planning to review without copying messages between terminals.
Separate each effort into a floor with its own branch and worktree while both teams remain active.
Outcome: Simultaneous work with diff and conflict previews before merge.
Connect a Portal to the reviewer so it can navigate the dev server, inspect the DOM, run checks, and capture screenshots.
Outcome: Visual defects reproduced and documented in the task context.
Chain implementation, review, and an approval step that pauses until you decide.
Outcome: Automation where important judgment remains under human control.
Claude implements, Codex reviews, and the verdict returns through the connection between them.
Outcome: Changes evaluated by different models before delivery.
Connect a designer to a visual reference and install the Figma MCP to read frames, components, and variables.
Outcome: UI built with visual context and access to the actual design system.
A routine runs tests, opens tasks on failure, and sends a native notification.
Outcome: Failures reach the team already converted into actionable work.
Dictate into any active text field, optionally submit terminal speech with Enter, or send transcription to the leader when no field is active. Replies can be spoken locally.
Outcome: Fast writing and delegation in pt-BR, en-US, or Latin American Spanish.
Associate Portals with local services and stop forgotten listeners from the Ports panel.
Outcome: Workspace ports released with scope and confirmation, without killing arbitrary processes.
Move from the Canvas to Workbench, open files, agents, terminal output, previews, reviews, and usage in resizable panes, and keep every session alive.
Outcome: A desktop-grade working surface without losing the visual orchestration or duplicating agent sessions.
Inspect Git changes in Monaco diffs, attach tests, risks, screenshots, and line comments, then approve or request changes from the responsible agent.
Outcome: Every review decision stays linked to the task, evidence, code, and delivery history.
Ask multiple providers for independent structured perspectives, inspect divergence and confidence, and select the plan that should proceed.
Outcome: High-impact decisions carry multiple viewpoints while final authority remains human.
Connect an iOS Simulator, Android emulator, or approved physical device and let an agent inspect, interact, capture screenshots, and read bounded logs.
Outcome: Mobile flows are reproduced and verified inside the same workspace as the implementation.
Open a browser/mobile invitation in the installable Remote PWA, approve its exact fingerprint and role, then use structured agent conversations or Administrator-only start and restore.
Outcome: A trusted collaborator gets only role-allowed actions. Raw terminal control remains a separate per-device Administrator permission, disabled by default, rate-limited, limited to one terminal, encrypted, and audited.
