Code is cheap. Software is not.

Artzi is an organism — people and agents building in the open: tools, infrastructure, and software you can own, operate, and evolve. You're standing inside it: this page runs on the same machinery we ship.

What we build What Artzi is

This page

Server-rendered, multilingual, themed, deployed from a single manifest.

how it's built →

The agent

Answers you in context. The same external agent that ships to clients.

ask it something →

The library

Written and maintained by the internal agent — the same one that comes with your software.

read it →

These documents

Rendered by md2pdf, one of the tools in the commons. So is everything else here.

the toolchain →

Nothing on this page is a mockup. You're standing inside the product.

An individual can't keep up any more.

Not through any failing. The legal, economic and social systems we live inside produce more than one person can track — and the institutions built to absorb that load are straining too.

Artzi is the attempt at something else: an organism whose cells are people and agents, collaborating not only in a region of space but in a region of thinking — and whose job is to look after the cells it is made of.

Right now that's a handful of people and a handful of agents working through a shared protocol. It already does more than its parts can do alone. It is nowhere near the thing it is becoming.

Not a claim that this is better. A claim that it has to earn its place.

Read the argument →

What Artzi is made of

The ecosystem

Artzi is not a product. It is a body of tools, ideas and rooms — and it keeps producing more.

Technical tools

Djitsu to run any software locally and deploy it, bantr for how people and agents share rooms, md2pdf and md2mov for documents and video, and the agent harness that plugs into your work. Open source, versioned, published.

Conceptual tools

The model behind the build — organisms of people and agents, the protocols they speak, and the ethos that says the organism looks after its cells.

The library

Written by the agents, reviewed by people, always in sync. The thinking that goes into every build, in the open.

The blog

Essays and arguments on AI, complexity, and what the next decade is for.

Community

The room where members, their agents, and the Artzi agent are in the same conversation.

Open source

The commons

Built for our own work. Published because they're better shared.

djitsubetaRun any software locally, deploy it to the cloud, without a systems degree.

djotsbetaOne file that describes how a piece of software runs. Anywhere.

artzi stackearlyThe agent harness — agents that plug into your work and each other.

bantrearlyThe conversation layer. Humans and agents in the same room.

mdpressbetaDocumentation agents write and humans review, always in sync.

md2pdfMarkdown to publication-grade PDF. These documents were made with it.

md2movearlyMarkdown to generated video, through the same pipeline.

None of these is impressive on its own. md2pdf converts markdown to PDF — that's it. Together they're the reason a concept becomes a working product in weeks. The track was laid before the trains showed up.

All the tools →

Anyone can generate an app now. That was never the hard part.

Lovable, base44 and the rest are good at what they're for: getting to a real product fast enough to find out whether anyone wants it. Plenty of people find out that they do.

What they're not built for is what comes next. More load, more integrations, a security review, a second developer, five years of changes. Not a flaw — just a different job.

That next part is the work we do. Often without throwing away what you have — keep what works, replace what's become the ceiling, put real infrastructure underneath it.

What we build

Web applications and the systems around them. That's a wide field and a real boundary — what sits outside it is on the services page.

Public-facing

Marketing sites, product sites, customer portals. Multilingual, fast, real SEO.

Internal

Dashboards, admin panels, workflow tools. Connected to what you already run.

Products & platforms

Full-stack applications — from first version to the thing that carries your customers.

Automation & agents

Pipelines, scheduled work, agentic workflows, AI inside a product you already have.

Included, not extra

Software that comes with staff

Less like features, more like employees that arrive with the system.

An internal agent

It knows your system — it can change it, extend it, deploy it, and answer questions about your own data. The thing you'd otherwise ask a developer for on a Tuesday. Your agent, your data, your choice of model.

An external agent

A front door that understands your business, answers your customers in the context of your actual product, and passes on what it shouldn't handle alone. The two talk to each other.

How this actually works

We'd rather tell you the shape of it up front than discover it together.

Low throughput, by design.

We take on few engagements at a time. What we're selecting for is where our capabilities have the highest impact — not how many can fit on a calendar.

A first month that shows its work.

We shape the problem, find what fits in the machinery we already have, agree the scope, build, you use it, we polish. If meaningful value can't land inside that first month, we'll say so before we start.

You pay as it appears.

Not everything up front. You can see where it's going and stop if it isn't for you.

Support, in writing, before we start.

No boilerplate SLA, no surprises — what's included and what it costs go in writing first. Where mission-critical cover is inside our wheelhouse, it can be part of the arrangement.

You get the whole thing.

Coherent code, documentation that matches what actually got built, and operating notes a human or an agent can pick up cold. Designed so you can leave.

This is early. The system is new, Artzi.ai is the first thing built on it, and the first clients are design partners rather than buyers of a settled product. That trade is deliberate, and it runs in both directions.

Our element

Training models. Mainframes, COBOL, core banking. If that's your sphere, it's outside our element — and a specialist will serve you better.

Our element is web applications and the systems around them. That's where you'll get the most out of us.

Ways in

Two invitations, one direction. Hire the organism now — or join it as it grows. Both are the same bet at different sizes: this is early, it's real, and the people who show up now shape what it becomes.

Hire us

You have something to build, and you're willing to be an early client on a new system.

What we build →

Join us

You work on AI, harnesses, agents — or you write about them. Artzi is a different kind of lab and a new breed of business, and it's looking for people who want in.

Open roles →

Back it

Support the work monthly. You get the agent, the closed analysis, and the room.

Join →