Skip to content
Fawaly by Daxow

FAWALY BY DAXOW · YOUR AI CODING CREW

Save your best agent for the big thinking.
Let the crew implement.

Keep Claude Code or Codex in charge of planning and review. Fawaly connects lower-cost coding assistants to handle suitable implementation tasks using the plans and models you choose.

Coming soon to the Mac App StoreWatch the 47-second tour ↓

Free app · macOS 14+ · Apple Silicon & Intel · Provider plans may cost extra

Not available for download yet. Windows and Linux are planned for later.

  • One visible Mac app
  • Your order, models & permissions
  • Review before accepting work

FAWALY / 00:47

Meet your new crew.

A quick look at the app: connect assistants, choose their order and follow a task through execution and review.

47 seconds · with sound · English interface · illustrative product showcase, not proof of your connection or balance

Read the video walkthrough
  1. Your primary agent plans; connected assistants implement.
  2. Choose an assistant and model, then set the team order.
  3. Connect coding plans, explicitly permitted APIs or local tools.
  4. Follow the active task and the assistant doing the work.
  5. Your primary agent inspects the result and records its review.
  6. Use a compact home screen, guided setup and Arabic or English.
  7. Fawaly is coming soon to the Mac App Store; Windows and Linux are upcoming.

STAY IN CLAUDE CODE OR CODEX

One request. A whole team behind it.

01

You ask

Describe the outcome in your existing coding agent.

02

Your agent plans

It sends a suitable, bounded implementation task through Fawaly.

03

The crew implements

The next eligible available assistant works within the approved scope.

04

Your agent reviews

It inspects the diff, reruns checks and records its verdict.

YOUR PRIORITY ORDER ACTUALLY MATTERS

First assistant busy? Let the next one help.

Set the assistant, model and effort for simple, complex and large-context tasks. For independent work in separate workspaces, Fawaly can select the next eligible idle assistant instead of treating one busy worker as a busy team.

An order change applies to new tasks. Conflicting workspaces or an entirely busy team wait or return a clear refusal. Paid fallback is never silently enabled.

Illustrative routing example

01Busy on another task
02Next eligible assistant
03Next in your order

Bring a plan you already use.

Guided setup helps you choose the right coding-plan, API or local-tool connection. Start with one assistant; you do not need every service.

  • MiniMax · M Plan
  • Kimi Code
  • Z.ai · GLM
  • DeepSeek
  • Ollama
  • LM Studio

Provider access, model availability and quota depend on your plan. A website chat subscription alone may not include coding or API access. New model suggestions only join your queue when you choose them.

A CLEAR FIRST RUN

Four steps to your first reviewed task.

01

Connect your main agent

Review the tool connection and delegation instructions for Claude Code or Codex, then open a fresh agent session.

02

Connect one assistant

Choose your service and plan. Follow guided sign-in or enter a key in the secure field, then check the connection.

03

Choose who goes first

Pick the model, effort and order for each task type. Keep metered fallback off unless you explicitly want it.

04

Try a small task

Ask your agent to delegate in a sample project. Match the task ID, assistant and model, then review the actual changes.

The app includes an optional card-by-card usage tour. You can also explore isolated Demo data without making real model calls.

See the four setup steps →

FAWALY / CHECK

How do I know Fawaly is working?

A connected badge is only the first check. Ask your agent to check Fawaly tools, delegate a small task, and show its task ID. The same assistant, model and task should appear in Tasks & reviews. Have the agent inspect the output and record a review. “Completed” means execution finished; it does not mean the work passed review.

A few things worth knowing.

Does installation make every agent chat delegate automatically?

Complete the agent connection and approve the delegation instructions first. Start a fresh session and verify the tools. Fawaly chooses the assistant when your agent delegates; it does not guarantee delegation on every message.

Does the app window have to stay open?

After setup, you can close the Fawaly interface. Its internal execution engine handles delegated work separately. Keep the coding-agent session open and your Mac awake and online during a task. Manage all settings from the single visible Fawaly app.

How much will I save?

The aim is to preserve expensive-agent usage for planning and review by moving suitable implementation to lower-cost assistants. Savings vary by task and provider; no percentage is guaranteed. Main-agent briefing and review still consume usage.

What permissions are needed?

Approve the connection and project scope you need. Keychain is used for provider keys; background activity may require macOS approval. Notifications are optional. Connecting assistants does not need Full Disk Access, screen recording or Accessibility control.

Where does my code go?

Keys use macOS Keychain and work logs stay on your Mac. Instructions and code included in an authorized task may be sent to your selected provider. A local log does not mean local AI inference; local tools depend on their configured endpoint.

Can I see balances and set limits?

Fawaly shows usage or quota where the provider supports it. An unavailable balance is not zero. Local limits can block future starts; they are not a provider-side billing reservation.

An idea worth crediting.

Ahmed Nagdy created the original project idea and the name “Fawaly”. We built on and extended his idea to create a native Mac app at Daxow. Our sincere thanks to him for the idea and the name.

Your agent leads. Your crew gets to work.

Coming soon to the Mac App Store. Explore the setup guide while we prepare the release.

Coming soon to the Mac App StorePrivacy & data