Some questions are worth decades.
Autonomio is an independent, non-profit AI research lab in Finland. Our work brings together language, mathematics, and the tools that help other people do science.
We have been following these questions since 2004. We fund the work ourselves, and give it the time it needs.
It began with language.
Beyond hand-written rules.
Our early natural language processing work introduced super-class summarization. It was an exploration of how a machine could generate content in a radically more efficient way than the conventional approaches at the time.
The question that arose from that work was how a machine might work with language without every possibility having to be anticipated by its maker. That question would stay with us.
Language as mathematics.
A decade later, we built an early prototype of a purely mathematical, relation-based approach to natural language processing using Wikipedia. Language was something to represent and work with mathematically, rather than handle through a growing collection of rules.
The prototype explored a direction that has remained central to our interests: what becomes possible when we reconsider the representation of a problem, rather than keep adding to the machinery around it?
Talos is one of our open-source tools for scientific researchers. It brought ergonomics to automating common scientific workflows to deep learning, allowing researchers to focus on what matters most to them, without compromising the full power of bespoke deep learning.
Thousands of researchers use Talos across scientific disciplines. We built it and kept maintaining it, including six years without a breaking bug. A tool used in science needs to be dependable long after its first release.
import talos
experiment = talos.Scan(
x=x,
y=y,
model=model,
params=params
)What we’re working on now
Making more science possible.
Our current work brings radical computational efficiency together with radical ergonomics: less unnecessary work for the machine, less friction for the researcher. We are developing tools that address practical barriers to breakthroughs across scientific fields.
Poise
Powerful tools. Less coordination.
Poise brings AI agents, software development, and long-form writing into one local workspace. We are working on making that capability easier to direct, so more attention can stay with the problem and less with managing the tools around it.
Terasweep
Explore more. Compute less.
Terasweep is our work on radically efficient experiment sweeps. By reusing calculations instead of repeating them, it aims to make larger spaces of possibilities practical to explore—not simply by adding more computing power, but by changing how much work is necessary.
Literview
A literature review that keeps working.
We are developing Literview for always-on literature review and semi-autonomous scientific discovery. It brings literature search, cited synthesis, data analysis, and follow-up questions into repeatable research cycles, with a record of the evidence behind the work.
Energy efficiency research
What does intelligence cost?
Getting a useful result matters. So does the energy it takes to get there.
Alongside our open-source tools, we are preparing a series of scientific papers on data center energy efficiency and machine learning energy efficiency. These are two connected areas of research: the infrastructure that runs computation, and the methods that determine how much computation is needed.
The question is not only how to do more with the resources available, but how much of the work is necessary in the first place. Our aim is to make more ambitious science possible without simply demanding more energy.
Research in progress. Papers will be linked here as they become public.
Independence, in practice.
Autonomio is a Finnish non-profit. We have paid for this work ourselves, without asking anyone else to fund it.
That gives us time to follow our own judgement. We can work on an idea before there is a market for it, share something because it is useful, and keep maintaining it after the initial excitement has passed.
Our work moves between language, mathematics, software, and the practical question of how to accelerate scientific research. We work closely with research groups at several universities, including Aalto University and Universidad Carlos III de Madrid (UC3M).
We started by playing with computers more than four decades ago, and never stopped.