Vision demo — illustrative, not literal

Science that runs itself

A shared digital twin of disease. An AI that finds gaps and designs discriminating experiments. Protocols that compile. Labs that execute. Results that update the twin — a closed loop whose only speed limit is physical time.

TA1Digital TwinTA2Science EngineTA3Protocol CompilerTA4Lab NetworkIGoRresearch loop
Mission control

Watch the organism think

Play, pause, or step the research pulse. Click a station to jump into its deep-dive.

Cycles
0
Experiments / day
0.0
Open gaps
4
Twin confidence
58%
live research logstreaming

Waiting for pulse…

TA1

Digital Twin

Interrogate a shared causal model that spans molecular → clinical. Hover a node — watch influence cascade across scales.

Scale focus
Molecular
Cellular
Tissue
Organ
Clinical
MetforminAMPKSGLT2InflammationGlucoseLipidsEndotheliumMyocardiumHbA1cBlood pressureMACEMortality

Interrogation readout

Metformin

Scale · molecular

Downstream influences

AMPKSGLT2GlucoseLipidsMyocardiumMACEMortalityHbA1cInflammationEndotheliumBlood pressure

Path belief (avg)

Belief-band thickness encodes confidence. Dashed red edges are gaps — places the Science Engine should look next.

TA2

Science Engine

Find the gaps. Rank hypotheses by expected value. Design the experiment that discriminates among mechanisms.

Gap radar
MetforminSGLT2SGLT2AMPK
  • Metformin → SGLT2 untested
  • SGLT2 → glucose path thin
  • SGLT2 → lipids missing RCT
  • AMPK → inflammation belief band wide

Reasoning trace

TA3

Protocol Compiler

A declarative experiment intent compiles into design, protocol, and instrument steps — then matches a capable lab.

1
Intent
2
Design
3
Protocol
4
Instrument
Lab capability match

Required

RCTlipidomicsecho

Boston-07

Boston

3/3

Oxford-03

Oxford

1/3

Singapore-01

Singapore

1/3

Zurich-04

Zurich

1/3

Tokyo-09

Tokyo

2/3

Cape-02

Cape Town

1/3
TA4

Lab Network

Validated robotic and cloud labs execute the protocol. Physical time is the only real latency — then model-ready data flies home.

TwinBoston-07SF-12Oxford-03Singapore-01Zurich-04Tokyo-09Cape-02São Paulo-05

Active node

Boston-07

Boston · n = 1,240

Physical experimentqueued
Model-ready packet

Waiting for assay completion…

Payoff

Closed loop

Results don't sit in a PDF — they update the twin. Belief bands tighten. The next cycle starts smarter.

Before

AMPKInflammation

0.42

belief

Prediction revised

cycle

After

AMPKInflammation

0.42

belief · band tighter

Impact

Velocity

Traditional science is a relay of years. The IGoR loop compresses it to the speed of physical experiment — humans steer, approve, and interpret.

Timeline compression

Today — handoffs~3 years
IGoR loopdays → weeks
twin → engine → protocol → lab → twin

Researcher in the loop: ask questions, approve designs, veto unsafe protocols, interpret surprises.

Idea → result

2–5 yearsweeks

Loop replaces handoffs

Gap → experiment

manualauto-ranked

Science Engine proposes

Protocol reuse

~0%>80%

Compiled, interoperable

Model update lag

yearshours

Results flow into twin

Wasted trials

highdiscriminating

Only experiments that resolve paths

Human role

do everythingsteer & approve

Ask, veto, interpret

This vision, grounded in what we already build

The twin, the intervene → predict → test → update loop, and living evidence are shipping today. The rest of the organism is the roadmap.

Illustrative fiction inspired by the IGoR program vision. Not a scientific claim.