The mind science fiction promised.
We're trying to build it
— in the open.
In every sci-fi film there is one: a brilliant, all-knowing mind that just
handles everything.
We want to actually build it — not for one giant company, but for every
individual, every organization, every city.
iDoris is that organizational brain, and it's also what we call ourselves.
Around it we do three things, and open-source all three.
Those three are the products we build. Day to day, they reach you as
four stages — Discover,
Enable,
Implement,
Operate.
You can stop after any one of them: Applied AI for real work.
iDoris was the AI's suggestion — a name you've seen a thousand times. By the time we went to register the domain, someone had taken it years ago.
So the main domain became iDoris.ai. Not a compromise — an improvement. iDoris is both what the org is called and what the core product is: the organizational brain. The work should be named after the brain anyway.
Our method gets us inside, iDoris trains models,
Agent24 & Hyphae connect reality and execute tasks
These aren't three parallel product lines. They're three preconditions for the same thing. Drop any one and the brain is just a demo.
How we workIn use
FDE — Forward Deployed Engineer — is Palantir's coinage for institutionalized grunt work: you have to get inside the customer to see the real jam. We think the role outlasts the name. It's the ANE, AI Native Engineer: not deployed, but present.
But "get inside" can't just be an attitude. We turned it into something reproducible — three sets a new hire can pick up and run with:
- A worked Discovery — all nine deliverables, filled in for real, zero placeholders
- A distributable interview skill — 3 guides, 36 questions verbatim, plus a self-checking scoring script
- Starter Kit designs — four components: Voice, Documents, Creative, Assistant
One test, and only one: can someone who starts tomorrow read it and get to work? If not, that's a bug in our documents, not in the person.
iDorisIn progress
A frontier model knows everything except your thing. iDoris tunes the model small enough, specific enough, close-fitting enough — until every individual, org, company, and eventually city has a brain of its own.
- Small-model tuning — LoRA, domain fine-tuning
- Industry training, vertical capability accrual
- Multimodality
- Multi-model collaboration and scheduling
Agent24 & HyphaeIn progress
A brain needs limbs. We use AI's different capabilities and different workflows to raise how well humans work and coordinate — a legion of agents under the brain's command.
- Agent development — memory, loops, context, skills
- Agent accounts & security — identity, authorization, blast radius
- Cross-org agent protocols — when every org has agents, how do they talk?
The future leads the present
If AI pushes the cost of productivity down toward the price of electricity, everyone has to answer one question: when my job is taken by AI, what's left for me to do?
What gets restructured is the work that exists purely in pursuit of extreme efficiency. The era needs people who collaborate with AI better, and it needs AI Native Engineers.
— on what keeps its valueWe believe three roles will emerge, and have already started emerging. They will continue to hold core value.
| Role | Why it survives | What we provide |
|---|---|---|
| 🗣 Expresser | AI will never say "this restaurant is pretty good" — because it isn't a person | One-click multi-platform publishing · illustration · de-AI-ing the prose |
| 💡 Innovator | When the whole map is fog, someone walks a path through it | Pain-point matching · experiment subjects · coordination |
| 🔧 Builder = ANE | The "small problem" you fix is what someone else endures every day | Model selection · LoRA · memory/context · agents |
The goal: build an open-source AI organization that contributes more digital public goods to humanity's digital future — and by doing so, raises the welfare of digital life itself.
— iDorisAI services from Thailand to Southeast Asia
Our long-term vision is an organizational brain for everyone. What we do today is practical — Applied AI for real work. Based in Chiang Mai and traveling to Bangkok, we help teams, communities, and organizations across Southeast Asia in four stages: discover what the work actually costs, enable the people who do it, implement what survives that scrutiny, and keep it working after launch.
Discover
We come in, measure how the work actually flows, and hand you nine deliverables — including the conclusion that you should not be doing AI yet.
Enable
Hands-on sessions built on what Discovery actually found — your examples, not a generic deck.
Implement
Four components already designed — Voice, Documents, Creative, Assistant. Wired to your process, not invented from scratch.
Operate
Launch is not the end. Measure, improve, expand — without this, the value of the first implementation decays within six months.
Discover → Enable → Implement → Operate. Enter wherever you are — and stop after any of them. Plenty of clients finish Discover and take it from there themselves; that is a successful delivery too.
![Jason [at] idoris.ai](assets/email-jason.png)
Ask together → research apart → come back with answers
Last session's questions come back as this session's answers. Questions ferment over time; answers grow out of practice.
Regular in-person gathering for AI builders. Zuzalu Library Event Space · Building F, 4Seas Nimman. Online option too.
🍄 Mycelium Protocol Blog — AI · Web3 · digital public goods. What we're thinking, what we tripped over, which calls turned out wrong.
Hands-on AI, step by step. Real practice on real problems — not a tutorial mill.
Name your pain
Tell us the jam you put up with every day. It might be the next thing a builder fixes.
Send a PR
Share your notes, your experiments, or finish a tool we started. Apache 2.0 — take it, fork it, no permission needed.
Intern in Chiang Mai
Chiang Mai University students: a paid internship — run hackathons, reach local companies, grow their AI-native capability.