A new kind of system. A new kind of AI model.
An operating system for organisations and communities, efficient by design.
See it work
The case for and the case against, from your own records, plus what’s missing from both, while it can still change your answer. When the result arrives the decision is scored against it, so you can show it held up, or learn that it didn’t.
Should our town adopt a 20mph speed limit?
For
I support the adoption of a 20mph speed limit within our town. Lowering the speed limit will dramatically enhance the safety of our streets for everyone. It reduces the severity of potential accidents, providing more time for reaction and a greater chance of survival for those involved. This is particularly vital for vulnerable road users like pedestrians, cyclists, and children. A 20mph limit promotes a calmer, more considerate environment. Evidence not supplied: data on accident rates before and after the introduction of 20mph zones.
Gemma 3 4B
Against
I oppose the adoption of a 20mph speed limit because it would likely result in increased traffic congestion and reduced efficiency for commuters. Evidence not supplied is needed to determine if the town's traffic volume and infrastructure can accommodate this lower speed. However, without such data, the proposal risks undermining the town's ability to manage traffic effectively and could lead to longer travel times and economic inefficiencies.
Qwen3 4B
Moderator
Disagreement: Proponents argue a 20mph limit enhances safety, while opponents claim it is unnecessary or impractical for traffic flow.
Unsupported: Each side makes broad claims without evidence, such as safety benefits or operational feasibility, which lack data support.
Needed data: Accident rates before and after implementation, public opinion surveys, and economic impact assessments. Evidence is not supplied.
Qwen3 4B
Composed from four model calls on the mesh. The debate engine is not deployed here.
Result: Insufficient evidence to determine outcome. The debate centers on safety versus traffic flow, but lacks supporting data. Crucially, evidence concerning accident rates, traffic volume, and public opinion is needed before a decision can be made.
Recorded on 19 November 2026 · replayed from storage
Use it
Your machines, your sites, your data. Nothing on someone else's cloud.
Datacentre OS →Show anyone the case you decided on, and how it turned out.
Intelligence Core →Everyone sees the same record, so nobody has to take your word for it.
Societal OS →Send a job, get the answer, and know which machine did the work.
Developers →WHERE WE ARE
Every change we merge lands here, next to the goal it’s working towards.
Left and right arrow keys move between changes. Home and End jump to the first and last.
Select a day to see its changes.
A change is one reviewed piece of work merged into our code: something new, a fix, a speed-up or routine upkeep. Each ball is one change. On busy days the balls are the most notable changes, and the bar beneath them counts the rest. Open any day to see every change. Counts show how active a project is, not how important it is.
When a column has more than eight changes, the balls lead with new work, then fixes, then speed-ups, then upkeep. Changes on the same date keep their server order.
One ball selects that change. The column background around the balls opens a finer period.
Tracking began on 17 November 2026.
Routing and serving for our own model infrastructure.
Reasoning and distillation behind Trigan models.
Community, governance and membership services.
Datasets we generate.
Work not yet assigned to a product.
Served from commit df264df · Every figure with its status
Join in
Free while we test. Every result tells you which model and machine produced it.
Not yet: adding your own machine as a node.