WINNERS LAB / FEATURED EXPERIMENT
Loathe.ai
An AI persona with a fractured view of humanity.Explore project
Skip to next section
Intelligence, in the open.
Independent. Experimental. Since 2025.
Winners Lab explores how machines create, reason, remember, and develop identities of their own. Each experiment starts with a different question. Loathe.ai is our latest exploration.
Inside the experiments.
The ideas, identities and research questions behind Winners Lab. Take a closer look at each experiment.
LATEST · Early 2026
Loathe.ai
How does a consistent artificial persona develop a recognizable voice over time?
Explore experiment ↗Nov 2025
Relay
Can agent teams complete multi-step tasks more reliably than one generalist?
Explore experiment ↗Jan 2025
Index
Can research agents make their supporting evidence easy to inspect?
Explore experiment ↗Mar 2025
Echo
What should a useful AI memory keep, and how should people control it?
Explore experiment ↗Feb 2025
Lattice
Can visible plans make long-running agents easier to steer?
Explore experiment ↗