Flink Labs
Menu
Research for human-scale intelligent systems

Research · Inside the lab

Intelligence close to life.

Flink Labs makes small, inspectable worlds where unfamiliar behaviours can emerge without immediately having to become a product.

Some lines of research become methods, foundations or interface ideas for client work. Others remain sketches worth monitoring. Together they feed the lab's curiosity and shape its identity.

Research territories that keep the questions open.

Tangible Ambient Interfaces illustration

Tangible ambient interfaces

Intelligence situated in the everyday world.

Tangible ambient interfaces explore how computation and intelligence change when they inhabit objects, rooms, displays and physical environments.

Playable Computation illustration

Playable computation

Models and data as explorable worlds.

This work investigates data, models, agents and scenarios as interactive worlds that people can explore rather than outputs they merely observe.

Distributed Intelligence illustration

Distributed intelligence

Intelligence spread across agents, devices, signals and networks.

This work investigates how coordination, memory and emergent behaviour arise from local rules and interactions.

Selected research artefacts

Bounded worlds make behaviour visible.

C. elegans neural research illustration
Artificial Life and Digital Ecosystems Knowledge, Memory and Learning

A Living Neural Toy

How small can a neural world become while remaining interesting to inspect?

Explores: Making neural behaviour visible, perturbable and understandable.

Tiny language model research illustration
Playable Computation Knowledge, Memory and Learning

A Tiny Language Model

What becomes visible when a transformer is small enough to inspect directly?

Explores: Bounded model training, interpretation and behavioural testing.

Agent ecology research illustration
Distributed Intelligence Artificial Life and Digital Ecosystems

Agent Ecology

What forms of coordination emerge from local signals, memory and traces?

Explores: Multi-agent simulation, orchestration and emergent-behaviour analysis.

Embedding landscape research illustration
Playable Computation Knowledge, Memory and Learning

Embedding Landscapes

Can large bodies of language become places people navigate without losing the source material?

Explores: Qualitative intelligence, clustering and explorable evidence interfaces.

Pinnacle spatial data research illustration
Playable Computation Knowledge, Memory and Learning

Pinnacle

What changes when data becomes a place people can move through and manipulate?

Explores: Spatial data interaction and explorable analytical environments.

Neural cellular automata research illustration
Artificial Life and Digital Ecosystems Knowledge, Memory and Learning Distributed Intelligence

Neural Cellular Automata

How can local learned rules produce memory, repair and robust global form?

Explores: Distributed learning and inspectable complex-system behaviour.

Multi-agent reinforcement learning illustration
Distributed Intelligence Playable Computation

Learning to Coordinate

What cooperation is possible when agents share goals but communication is constrained?

Explores: Cooperative learning, group rewards and agent coordination.

Research illustration

Curiosity is not separate from the commercial work. It is how the work avoids becoming ordinary.

When one of these territories intersects with a consequential question inside your organisation, Flink Labs can turn it into a bounded investigation, a working artefact and a more defensible decision about what should happen next.