People-centered cooperative
Specialized agents that work for your business
We build AI agents that get to work on your operation every day, around the clock. They are designed to extend your team, not replace it: real people set the goals, review every action and answer for each agent.
01 — Agents as a service
Digital specialists, built to fit
We don't sell generic software: we create agents. Each agent is a digital specialist designed for one concrete task of your operation, executed with bounded autonomy: it works on its own, but a person reviews the important decisions.
01 One specialist per task
Each agent is defined to solve one specific problem of your business, with clear goals and an agreed way of working.
02 They work on your real operation
Agents operate the systems and tools your business already uses: they fit your way of working, they don't ask you to change it.
03 People in charge
Behind every agent there is a human team from the cooperative that designs it, sets its limits, tunes it and responds when something changes. Technology executes; people decide.
02 — Our main agent
MAGI runs your operation, non-stop
MAGI is our main agent: it works around the clock, autonomously, within agreed limits, and in the background, operating your business systems without depending on a screen or someone watching over it.
01 Always on
MAGI doesn't sleep or take holidays: it monitors and executes your operation tasks at any hour of the day.
02 Headless operation
It executes directly on your systems, without constant intervention: you set the goals and MAGI takes care of the work in the background.
03 Human decisions
We set its goals, review its actions and tune it with you. The final word is always a person's.
MAGI agent network in motion
MAGI network at rest
[magi] 09:41 — daily operation running [magi] 09:41 — billing synchronized [magi] 09:42 — order #1042 processed [magi] 09:42 — report sent to the team
03 — Hardware for agents
Devices to control and talk with your agents
We are designing physical objects so your agents don't live only inside a computer: simple devices to give them orders, listen to them and watch them work.
01 Orders without a keyboard
A button, a dial, your voice: the same intention, another surface. The agent understands either way.
02 Your operation, in plain words
Devices that tell you what your agents are doing in the language of your business, without endless dashboards.
03 Concept in development
We are prototyping this line together with our community. No promised dates: when it is ready, you will read about it here.
In development
We are exploring device prototypes together with our community. Nothing is open for enrollment and no dates are promised: when something concrete exists, you will read about it here.
04 — Ecosystem
A cooperative, a community
Joyan is a collective project: people who create agents, run them and share what they learn. We believe AI is worth it when its benefits are shared: the more businesses adopt it with judgment, the stronger every agent and every community.
01 Cooperative
We are people who cooperate: knowledge, tools and results are shared, not concentrated.
02 Collective creation
We design and operate agents together with each business; what one project teaches improves the next one.
03 Made by people
No smoke and mirrors: a human team designs, operates and improves every agent from real feedback from real businesses.
In development
The community is taking shape
We want more people to create, operate and share agents — and to adopt them with judgment: knowing what they can do, what they can't, and what is worth reviewing in person. Today we are putting together guides, content and training, and looking for first partners to build with us. None of these are an enrollment, a form or a delivery offer yet: they are paths under development that we will open right here.
Small businesses
Guides and content to adopt agents with judgment: what they can do, what they can't, and what is worth deciding in person. No tech jargon.
Development teams
Training and materials for teams that want to build their own agents with judgment and autonomy.
01 Guides and content
We are writing materials that explain how we design and operate agents, aimed at small businesses and development teams.
02 Training
We are designing training so more people can learn to create and operate specialized agents.
03 Agent beta participation
We are looking for businesses that want to build agents with us in an early stage. It is still in development: when we open slots, we will announce it here, without promising availability.
04 Hardware-agent prototypes
We invite conversations about device prototypes for agents. This is ongoing exploration: no dates, no enrollment and no delivery.
05 — Log
Our progress, in our own words
We record here what we are building: prototypes, integrations and experiments, with what works and what we are still unblocking. No promises: just the step by step.
-
[Research] Exploring
Testing surfaces for giving agents orders
We are testing buttons, dials and voice commands as simple ways to give an agent orders. We are learning which surfaces feel natural in a real operation and which do not yet.
Read the entry -
[Software] In progress
The action log of every agent
We are designing the log where every agent leaves a record of what it did, so a person can review it at any time. Without a log there is no human review: it is the foundation of everything we build.
Read the entry