Three products.
Built by the tool
we're building.

MOOOON is building three pieces of software — a task manager, an AI-team desktop runtime, and the project-management tool that plans them both. The third one isn’t shipped yet. We’re using its workflow to build the other two anyway. See how.

Scroll
01 · Taq

An app for your time. A desktop for your AI team.

Two apps under one brand. Taq is where your tasks, calendar, and groups live. Taq Hub is where your AI agents work. They're built to play well together — but each one stands on its own.

01a Coming soon

Taq

All-in-one time and task management. Tell it what you want; it makes the plan.

iOS · Android · Web

Taq holds your to-do list, your calendar, and the shared lists and calendars of every group you're in — your family, your team, your friends planning a trip. One surface for everything time-shaped.

The interesting part: an AI planner built in. Tell it "learn guitar in six weeks," "train for a 10K," "launch my side project this quarter," "plan my sister's birthday party" — it figures out what kind of plan you're asking for and lays it out as real, paced tasks on your calendar, not a wall of generic advice.

9:41
Taq AI Planner
Tell me what you want to do. I’ll make a plan.
Learn guitar in 6 weeks|
Planner
● Got it — a 6-week learning plan.
I’ll stagger the practice, add a few video resources, and ramp up slowly.
Learn guitar in 6 weeks
Your 6-week plan · 18 tasks
● Tune the guitar Mon · 15m
● First chord: G major Tue · 30m
● Rest day · listen to playlist Wed
● Second chord: C major Thu · 30m
● Switch between G ↔ C Sat · 45m
● Week 2 · D major + first song …
Tell Taq what you want. Learn a skill, train for a race, build a side project, plan an event, lock in a daily routine — it figures out what kind of plan you need and paces it like one.

Things the planner is good at: learning a skill, daily routines, health & fitness, creative projects, building a habit, events & logistics. Each gets its own pacing, its own questions to clarify the brief, its own resource style — because how you learn guitar isn't how you train for a marathon.

When you connect them

The app is the remote.
The desktop does the work.

Taq can hand a task to a Taq Hub team. Connect them once — like any other integration — then assign work from your phone, watch the agents run on your Mac, get the result back when they’re done.

9:41
Taq Tasks
Today
Draft launch post
Reply to mom
Pick up groceries
Draft launch post
Assign to →
Taq Hub
Content Creator team
✓ Sent to Hub
Draft launch post
Content Creator team · working
Researching… 25%
● Hub now
Draft is ready
280 chars · 1.2k article · scheduled to X, LinkedIn, Threads
Review →
taq-hub MLX · Local
Content Creator team · idle Content Creator team · running
Researcher
Copywriter
Editor
Social
→ Received from Taq
→ Researcher · 5 sources pulled
→ Copywriter · 280-char post drafted
→ Editor · tone pass, fact-check passed
✓ Social · scheduled X 9a · LI 10a · Threads 11a · workflow done in 41s

Connect once — like any other integration — then assign from the app, watch Hub run it on your Mac.

01b Coming soon

Taq Hub

A desktop harness for local and remote AI. Run agent teams on your machine.

macOS · Apple Silicon · MLX

Taq Hub isn't one assistant — it's a desktop home for teams of agents that work together on a workflow. Start with a built-in team from Taq (Content Creator, Brand Designer, Product Designer, E-commerce Seller) or build your own — design the agents, pick their tools, write the workflow that connects them. When it's good, publish it to the marketplace.

taq-hub · Content Creator team MLX · M2 Max · 14B Qwen
Goal
Publish a launch post about MLX speedups across X, LinkedIn, Threads.
Researcher
web search
+ MLX
Copywriter
draft post
via MLX
Editor
tone +
accuracy
Social Mgr
schedule to
X · LI · Threads
→ Researcher · pulled 5 MLX benchmark sources
→ Copywriter · drafted 280-char post, 1200-char article
→ Editor · tone pass, fact-check passed
→ Social Mgr · scheduled to X 9am, LI 10am, Threads 11am
✓ Workflow complete · all on-device · 41s total
A team of agents, one workflow. Hub runs them locally with MLX on your Mac. Pick a team from Taq, build your own, or hire one from the marketplace.

On macOS, Hub runs local models through Apple's MLX framework — the unified-memory, Apple-Silicon-native runtime that makes 7-70B-parameter models genuinely fast on a MacBook, no GPU rental required. Cloud LLMs work too. The headline is on-device: your code, your keys, your repo, your hardware.

02 Coming soon

Prodect

AI-native project management. Open core.

{ } Open core · GPL-3.0 + hosted brain

Open-source Jira, with planning intelligence built in. Describe what you want to build; Prodect produces an Epic → Story → Subtask tree, hands the coding Subtasks to your coding agent (Claude Code, Cursor, Devin, Aider) with a rich prompt it can execute end-to-end, and leaves the rest for you to do yourself. It's using its own workflow to build itself — and Taq.

prodect · my-side-project Pre-plan phase Build phase
Tell Prodect what you want to build
I want to build a Notion-like notes app, mobile-first, with end-to-end encryption.
Good. A few clarifiers first — solo or with collaborators? Web app too? Want me to suggest a tech stack?
Solo for v1. iOS first. Surprise me with the stack.
Updated vision.md, stack.md, workflow.md. Ready when you are.
Produced reports
md vision.md just now
md stack.md just now
md workflow.md just now
↻ iteration loop: docs update as you and Prodect refine the idea.
Epic → Story → Subtask
▼ Epic 1 · Auth & sync
▼ Story 1.1 · iCloud account scaffold
• 1.1.1 Set up Apple sign-in code · 12m
• 1.1.2 CloudKit container provisioning code · 18m
• 1.1.3 Privacy policy draft legal · 30m
▸ Story 1.2 · E2EE handshake
▸ Epic 2 · Notes editor
▸ Epic 3 · Search
Executor
Subtask 1.1.1
Set up Apple sign-in
→ Claude Code
$ git checkout -b 1.1.1-apple-signin
Added AuthenticationServices entitlement
Wrote SignInWithAppleButton view
Wrote AuthService.handleAuthorization
Test passing: signs in, persists token
✓ PR #12 opened
Subtask 1.1.1 done. Pick next →
Two phases, one loop. Chat with Prodect to shape the idea. When you’re ready, it produces the tree and hands Subtasks to your coding agent one at a time.

prodect-core is GPL-3.0 open-source — Work Items, Stories, dependency graphs, multi-tenant workspaces, the tree-view UI, GitHub integration. The planning brain (prodect-ai) ships as a hosted service. Browsers only ever talk to prodect-core; the AI layer is server-to-server. Same playbook GitLab, Sentry, Plane, and Mattermost use.