Think freely, on your Mac.
Local inference has no external LLM API token charges. Use your Mac's compute for drafts and iteration.
LOCAL-FIRST AI GENERATION OS
Your Mac.
Your AI workspace.
Research, write, build. Bring on-device AI, conversations, creative tools, and memory together. Connect external agents when you need them, and take your work through to an artifact.
Apple Silicon · macOS 14+MLX / GGUFExternal connections are optional
DENOJU / macOSAPPLE SILICON NATIVE
Real app screens · reference images from the time of capture
Features reflect development as of Oct 2, 2026. The downloadable beta is 0.5.6 / build 21; available features and screens may differ.
01 / WHY DENOJU
More freedom. More ownership.
Keep creating in the conversation.
From a question to a document, code, or website. Instructions, progress, and artifacts stay connected.
Keep what worked.
Knowledge in Memory. Procedures in Skills. Standards in Rules. Make good work reusable.
02 / FROM CHAT TO ARTIFACT
One conversation. Many possibilities.
No separate AI composer for every tool. Addons hold settings and artifacts; your requests start in the shared chat.
DOCS
Documents & presentations
Create PDF, Word, Excel, and PowerPoint files from your brief.
CODE
Code & development
Select a project. Ask for investigation, edits, builds, and tests.
WEB
Websites
Plan, design, generate, and edit a static site. Check it in preview.
MEETING
Meetings & roles
Bring people and multiple AIs around a goal. Assign planning, building, and review.
MEDIA
Images, video & music
Create media with compatible models and additional runtimes.
Models and runtimes requiredAUTOMATION
Mac actions & automation
Operate allowed apps and repeat tasks within permissions, approvals, and stop conditions.
03 / HOW IT WORKS
From an idea to something you can use.
A proposal, for example.
- 01
Ask in your own words
“Make a three-section Word proposal for our new service, aimed at small businesses.”
- 02
Connect the context and tools
Your selected model works through the brief. The Host checks document-tool availability and runs the task.
- 03
Review and keep talking
Open the file and ask, “Make section two shorter.” The artifact stays connected to its conversation.
THE DENOJU DIFFERENCE
Beyond running a model.
DENOJU connects AI, work context, and creative tools. Your existing models and external agents can become part of the workspace.
Models + a workflow
Connect MLX/GGUF management with chat, projects, documents, websites, and meetings.
Local + external intelligence
Start with on-device AI, connect external agents when needed, and choose which past context to share.
Today's artifact + tomorrow's foundation
Reuse knowledge, procedures, standards, evaluations, and operating settings for the way you work.
04 / YOUR LOCAL INTELLIGENCE
Choose AI that fits your Mac.
MLX and GGUF support. Discover models on Hugging Face or use compatible LM Studio and Ollama libraries. See where a model lives and whether it can actually run.
Hugging FaceLM StudioOllama
YOUR MACMLX · GGUF
- On this Mac01
- Installable catalog02
- External search results03
- Cloud / API04
- Reference only05
05 / KEEP WHAT WORKS
Good work today. A better starting point tomorrow.
Keep your judgment and methods in reusable form. Growth means accumulating knowledge and workflows; conversations alone do not automatically retrain a foundation model.
01Memory
Knowledge to keep
02Skill
Steps to reuse
03Rule
Standards to follow
04Profile
Settings tested
GET STARTED
Start in three steps.
- 01
Install on your Mac
Apple Silicon and macOS 14 or later. Follow the app's setup and account instructions.
- 02
Choose a model
Install and load a model that fits your memory. Compatible existing libraries work too.
- 03
Ask for your first task
Start in chat. For file-based work, select your project and output location.
Does everything work offline?
Local inference with an installed compatible model runs on your Mac. Model downloads, accounts, web search, external agents, and shared features can require network access.
Are external AIs free to use?
Local inference has no external LLM API token charges. External agents and cloud providers apply their own subscription or API fees.
Can I use my existing models and agents?
Compatible MLX/GGUF files and LM Studio or Ollama libraries can be used. External agents need a supported CLI or connection and authentication. A search result is not a guarantee of compatibility.
Can I generate media immediately?
Compatible models and additional runtimes are required. Available routes depend on your Mac's memory, formats, and connections.
Can the AI act or publish without asking?
The Host validates capabilities and permissions. Actions requiring approval wait for confirmation. OS access does not grant unrestricted sending, deletion, purchasing, or publishing.
Your work. Your AI.
Think locally. Create through conversation. Keep what works. Start on your Mac.
Try Mac Beta for Free