Orchestrate your coding agents across every project

Orchestrating coding agents means running more than one AI coding agent at the same time — and giving yourself a place to launch, watch, and steer all of them without losing the thread. A single agent, like Claude Code, Codex, Cursor, Grok CLI, or OpenCode, is great at planning and executing a multi-step change inside one repository. Orchestration is the layer above that: which agent is working on which project, what each one changed, which task is next, and how to keep several of them productive at once.

Aiola is that layer. It's a local desktop app that orchestrates Claude Code, Codex, Cursor, Grok CLI, and OpenCode across every project you run. It is not an editor or an IDE, and it is not a model provider — it never resells compute or marks up an API key. It's the operations layer that sits on top of the agents you already pay for and turns a pile of terminal windows into one command center.

Concretely, orchestration answers three questions at once. First, dispatch: which agent should take this task, and in which project? Second, isolation: how do you keep three or four agents working at the same time without their changes colliding? Third, continuity: how does work carry on when you step away — on a schedule, from a task queue, or through an agent that already knows the codebase? A terminal can run one agent well. It can't answer those three questions for a whole portfolio of projects. That's the gap an orchestrator fills.

The moment you run coding agents in more than one repo, the hard part stops being the code and starts being coordination. Which project had which agent session? What did Codex change yesterday in the other app? Where's the task you meant to hand off before you closed the laptop? With bare terminals, all of that state lives in your head and across a dozen tabs — and every context switch costs you.

Orchestration solves the coordination problem:

• One window shows every project's active threads, recent changes, and pending tasks.
• Each agent runs in an isolated thread, so a fix in one repo never bleeds into another.
• Tasks and automations keep agents working on a schedule instead of only when you're watching the terminal.
• You choose the right agent for each job instead of being locked into one vendor.

A single-vendor GUI can only orchestrate one provider. A terminal multiplexer like tmux gives you parallel shells but no operations layer and no central view. Orchestration means both: many agents, one place to run them.

Aiola gives orchestration a set of concrete surfaces:

• Threads — each thread is an independent agent session with chat, diffs, branches, a terminal, and a built-in browser. Run several in parallel and place them side by side in split view or on a canvas whiteboard.
• Tasks that trigger agents — a per-project board where a task can launch a coding agent and hand the result back for review, so work moves without you babysitting a prompt.
• Automations — recurring scheduled agent jobs (triage, checks, summaries) with reusable templates and a run history, so agents keep working on a cadence you set.
• Agents — reusable agent definitions with their own role, skills, instructions, and memory, so an agent shows up already knowing how a given project works.
• A central cross-project view that lets you steer the whole portfolio at once — plus app logs, analytics, and feedback living next to the code instead of in five other tools.

Pick the provider per thread or per task, and switch depending on the job. Because everything shares one workspace, an agent's output — a diff, a branch, a task result — is right there for the next agent or for you to review, instead of being trapped in a terminal scrollback you already closed. That's orchestration you can actually see and control.

Most agent GUIs manage exactly one provider. Aiola orchestrates Claude Code, Codex, Cursor, Grok CLI, and OpenCode from a single app, on your own subscriptions. That matters because no single agent wins every task — one is better at a sprawling refactor, another at a fast fix, another at exploring an unfamiliar codebase. When you can reach for any of them from the same window, you stop shoehorning every job into whichever tool your GUI happens to support.

Cross-vendor reach is the difference between a single-tool wrapper and a real orchestrator. It's also durable: as new agent CLIs mature, an orchestration layer that already speaks multiple vendors is where they plug in.

Aiola runs every agent on your own machine, against your own repositories, using the Claude Code, Codex, Cursor, or Grok CLI subscription you already have. Your code is never uploaded to Aiola — it stays on your machine and goes only to the AI provider you already subscribe to — there's no cloud runner, no compute resale, and no API-key middleman taking a cut. Threads, tasks, notes, and automations are stored locally. Optional analytics, logs, and feedback telemetry is ingested through Aiola's cloud.

That's a deliberate design choice, not a limitation: orchestration should give you more control over your code and your agents, not hand them to someone else's server. Read more about how Aiola keeps your data yours.

Aiola runs on macOS 11+ (Apple Silicon and Intel) and Windows 10+ (64-bit). If you're on Windows, that matters: several agent GUIs are macOS-only, so cross-vendor orchestration on Windows is hard to find. A phone companion is coming. When you're ready, just download Aiola and point it at your projects.