Autonomous Unit Intelligence

A Collective Intelligence for Autonomous Systems

We build shared AI systems that see together, learn together, and act together — enabling scalable autonomy across complex environments.

01 / THE PROBLEM

EVERY UNIT
IS AN ISLAND

Today's autonomous systems are born alone. Each unit perceives the world through its own sensors, makes decisions in isolation, and dies with everything it learned. No transfer. No coordination. No shared memory.

02 / THE CEILING

BILLIONS SPENT.
THE CEILING HOLDS.

Hardware has caught up. Compute is cheap. Sensors are precise. Yet autonomous units still fail in the long tail — the rare, ambiguous, critical moments that isolated experience alone can never prepare for.

03 / THE SHIFT

THE PROBLEM
ISN'T THE UNIT

Intelligence that doesn't compound is automation. What autonomous systems lack isn't compute — it's collective memory. A shared layer where every unit learns from every kilometre driven by every other. That's what Ithryn builds.

01 / 04

Perceive

Sense everything.
Miss nothing.

Our multimodal perception stack fuses LiDAR, radar, camera and V2X signals into a unified world model — refreshed and shared instantly across the entire fleet.

02 / 04

Connect

One fleet.
One mind.

The Ithryn mesh layer connects every unit into a live shared environment — a world model no single unit could construct alone, continuously updated in real time.

03 / 04

Decide

Reason at
machine speed.

Coordinated inference across the fleet — weighing thousands of variables in parallel, drawing on collective memory trained across every kilometre ever driven.

04 / 04

Improve

Every kilometre
compounds.

The shared model learns from every journey driven by every unit, simultaneously. Every new unit accelerates the system. Intelligence that scales with the fleet.

scroll

01 / 03

Perception

See everything.
Miss nothing.

Our multimodal perception stack fuses LiDAR, radar, camera, and V2X signals into a unified world model. Every unit in the network sees what every other unit sees.

SENSOR FUSION //

02 / 03

Decision

Not rules.
Real intelligence.

Ithryn's onboard AI reasons continuously — weighing thousands of variables in parallel, drawing on a shared model trained across the entire fleet. No scripted decision trees. No central bottlenecks.

DECISION ENGINE // ACTIVE

03 / 03

Network

Units that
cooperate.

The Ithryn mesh protocol lets units share intent, trajectory, and sensor data in real time — forming a collective intelligence no single unit could achieve alone.

MESH PROTOCOL // NOMINAL

Timing

Three forces.
One inflection point.

01

Hardware is ready.

Edge chips like NVIDIA Orin and Qualcomm Ride now deliver the compute needed to run distributed AI onboard, in real time. This wasn't possible three years ago.

02

Regulation demands it.

UNECE WP.29 and EU type-approval frameworks for autonomous units take effect 2025–2026. OEMs need certifiable, auditable AI systems — now.

03

Isolated AI has hit its ceiling.

Waymo, Cruise, Mobileye — billions spent on per-unit intelligence that doesn't scale. The industry knows the approach is broken. No one has built the collective layer. Until now.

Get in touch

Let's talk.

Whether you're an investor, a potential partner, or an exceptional person looking for a defining problem to work on — we want to hear from you.

anik@uddin.se