Concepts

A handful of ideas explain how Maestrly works. Once these click, everything else follows.

Workspace

A workspace maps a project to a local Git repository. It's the home for the conversations, boards, notes and memory of that project. Adding a workspace doesn't move or copy your code — Maestrly works in place, in your repository.

Conversation

A conversation is a working session with an AI agent inside a workspace. Each conversation is born on its own isolated branch and worktree, so multiple agents can run in parallel without stepping on each other. You choose which CLI handles a conversation when you create it, and you can switch tools later (see Handoff).

CLI (your AI coding tool)

A CLI is the AI coding command-line tool that actually does the work — you bring your own. Maestrly detects the tools installed on your machine and runs them for you, showing each tool's terminal UI inside the app. Maestrly orchestrates the tools; it never resells or proxies your AI usage, and your subscription stays with the provider.

Handoff

Handoff lets you switch the CLI handling a conversation without losing context. When you change tools mid-conversation, Maestrly packages what was said and done so far and hands it to the new tool to read before it continues — so the new agent picks up where the previous one left off. This lets you play to each tool's strengths within a single thread of work.

Board (Kanban)

Each workspace has a Kanban board. Beyond organizing tasks, columns can be configured as "agent" columns that automatically dispatch a conversation when a card lands — turning the board into a pipeline (for example: produce → review → done). Cards and conversations are linked, so you can jump from a card to the agent working on it and back.

Notes, memory and delivery search

These three look similar but serve different purposes — Maestrly keeps them distinct:

• Notes — multi-page, Notion-style notes attached to a conversation or to the whole project. You write them, and the agents can read and write them too. Good for specs, checklists and scratch context.
• Project memory — a per-project memory that Maestrly injects into the AI tools when they start, so durable facts about the project survive across conversations and tools. It's separate from notes: memory is the small, always-on context; notes are the longer documents.
• Delivery search — when work reaches the board's "Done" column, its delivery is indexed locally so later conversations can search past deliveries for relevant context. It's retrieval over what you've already shipped, computed on your machine.

Local by design

Across all of these, the rule is the same: the AI tools and your credentials run and are stored locally, in your OS keychain. Your code, prompts and the agents' responses are not sent to Maestrly's servers — only the minimum needed to run your account.

Next: Features overview.