How much model does the task need?
Retrieved-Span Training for Efficient Query-Focused Meeting Summarization on QMSum.
Read the paperSovereign AI for every product.
Ertas AI is an applied AI lab in Melbourne. We're building the system for self-improving models.
A model for every job.
Train a model for each high-volume task, test it on your evals, and ship it when it beats your current API on cost and latency.
How we work with enterprise teamsFrontier-level results,
right-sized specialists.
Measured in our study on a summarization benchmark (QMSum).
Read the paper- 99%
lower cost than Claude Opus 5
- +18%
better than baseline
- 23×
faster than Claude Opus 5
What makes it sovereign.
Train a model on your data. Run it in the cloud or on-prem. Continual improvement loop by invitation.
Explore the endpoint programme
Our research
All researchModels that learn from traffic
How production requests and feedback become training data, and how a new version earns its place.
Agentic unit economics
Where AI spend goes, and what it costs to run agents at scale.
Read the analysis