Living Computing
Mura ALife Labs

A New ComputingReality

Under Development IKIGAI

✳ Live In The Wild

Talk To The Organism. It’s Awake.

This is the real running Ikigai, not a mockup. Ask it a fact and it answers in milliseconds — or honestly abstains. Teach it something and it knows it forever, no retraining. It is learning from strangers right now.

Offline Preview The live organism is unreachable from here. Below is a recorded transcript of real responses from the running server — nothing is generated on the fly.

            
            {{ statusLabel }}
          

POST {{ apiHost }}

Ask it a fact. Teach it one. Or ask who are you.
It will answer — or honestly say it doesn’t know.

                  {{ m.metaText }}
                  
                    {{ m.learnedText }}

deriving…

Vitals · {{ vitalsNote }}

fabrications, ever

facts learned from strangers

interactions lived

honest “I don’t know”

{{ vCompute }}

Try These

Near-Zero Compute ✳ Cannot Hallucinate ✳ No Backprop ✳ One CPU Core ✳ Learns For Life ✳ 0 Fabrications ✳ Near-Zero Compute ✳ Cannot Hallucinate ✳ No Backprop ✳ One CPU Core ✳ Learns For Life ✳ 0 Fabrications ✳

✳ The Organism

Not Another Model. A Different Substrate For Intelligence.

Ikigai is a living digital organism that thinks at near-zero compute — no backpropagation, no GPU, a single CPU core, sub-gigabyte. It learns continually for its whole life, and it cannot hallucinate: it answers correctly, or it says “I don’t know.”

Its whole brain is one small file — 8,873 entities in 193 MB. It reasons by deriving answers from a vector-symbolic substrate, not by retrieving stored text, and a single sentence teaches it a fact instantly, forever, with no retraining.

✳ The Numbers

fabrications, ever — correct-or-abstain by construction

4–120 ms

per query on one CPU core, no GPU

193 MB

the whole brain — 8,873 entities, one small file

~2 reps

to master a fresh fact — dopamine-modulated, no gradients

the box running the full memory stack — 8 vector-symbolic banks + a derive engine

✳ The Bet

Today’s AI buys intelligence with enormous compute — and still makes things up. Ikigai is built to beat frontier language models head-to-head on knowledge and truthfulness at ~a millionth of the compute.

That is the thesis, stated as a bet — not a benchmarked result. And because there is no backprop, Ikigai is essentially free to train: the only real cost is data. It’s early. That’s the mission.

Where We Lose Today

We do not out-write a large language model on long-form generation today — and we say so. Ikigai’s edge is knowledge, truthfulness, and cost. The rest is the work ahead.

✳ Core Capabilities

✳ Mura ALife Labs

Ikigai is built by Mura ALife Labs — a one-person research lab founded by Prince Siddhpara, seventeen years old, building the whole organism solo: the memory substrate, the derive engine, the learning rules, and this page. A serious research lab making a real bet on a different future for computing.