๐Ÿ–ฅ๏ธ Desktop & web

The AI project studio
that ships real work

Talk to a coordinator that knows your whole project. It interviews you, writes the plan, generates the tasks, and runs a team of agents โ€” with Git, builds, and metering built in.

๐ŸŽ™๏ธ Voice or text ๐Ÿค– Multi-agent ๐Ÿ”€ Routed & metered
Nexus Projects setup board, scoped by Client and Project
โฌ‡๏ธ Download

Get the Nexus Projects desktop app

Free desktop client for Windows, macOS, and Linux. Always the latest release.

๐ŸชŸ
Windows

.exe installer ยท 64-bit

Download for Windows
๐ŸŽ
macOS

.pkg ยท signed & notarized

Download for macOS
๐Ÿง
Linux

.deb package or portable .tar.gz

All downloads & release notes โ†’

The Coordinator interview
Setup interview

Turn a conversation into a plan

The Coordinator interviews you with short, multiple-choice questions and proposes tags onto the board as you talk โ€” industries, platforms, objectives, features, languages, frameworks, and libraries. Answer by typing, or go fully hands-free: real Silero VAD detects end-of-speech and OS-level echo cancellation lets you talk to it like a phone call.

The plan editor
Living plans

Plans every layer shares

Generated plans are living Markdown documents โ€” Overview, Client, Server, Database โ€” that become the project-wide context every agent reads. Pick which Coordinator agent owns each plan, and edit freely as the project evolves.

Task detail with tabs
Tasks

Hierarchical tasks, deep detail

Plans drive hierarchical task generation. Every task has Overview, Sub-Tasks, Git Changes, Agent Work, and Builds & CI tabs โ€” status, priority, dates, and assigned agent all persist live and reflect instantly across the app.

Agents and personas
Agents

A team of specialized agents

Project Manager, Coordinator, and SDE specialists โ€” Generalist, Networking, Physics, Database, UI/UX, and DevOps โ€” plus a Verification agent. Each persona is backed by a configurable model, with its own prompts and tool permissions.

AI providers
Inference

Bring your own inference

Automatic LAN discovery finds Lemonade inference servers on your network, alongside the providers you configure. Pick exactly which model runs each agent โ€” local box, cloud GPU, or the metered Nexus gateway.

Code and Git workspace
Code & Git

Git, built in

An integrated workspace with full Git โ€” commit, push, pull, merge, branch, and history โ€” plus an editor and source-control view over your project's plans and code. The agents work where you work.

Launch and packaging
Launch

Build & package from the app

Docker-backed build environments, images, and containers โ€” with live build logs โ€” so you can package and ship without leaving the studio.

Account and usage metering
Account & usage

Metered, never cut off

Usage is metered against your plan for the billing period. Go over and output is simply paced โ€” never hard-blocked โ€” with an upgrade prompt, so the app keeps working. Plans run from Basic to Platinum with optional add-ons.

Put your project's main brain to work

Create an account, point it at your inference, and start shipping.