Case studies.
The problems, experiments, and outcomes behind our customer collaborations and independent research.
3 case studies
- Evals
- Self-improvement
DetectBench: evaluating investigative agents with Optimus
Working with Resemble AI, we built DetectBench to measure investigative agents and improve them. Optimus implemented the evaluation infrastructure, investigated failures, and tested successive agent revisions.
case study · AUG 30, 2026
- Fine-tuning
- RLVR
Gemma: teaching a language model to paint with Optimus
With Google DeepMind, we explored teaching Gemma to paint through code. Optimus built the rendering and evaluation pipeline for supervised fine-tuning and reinforcement learning.
case study · AUG 29, 2026
- Robotics
- Teleoperation
MuJoCo: building a robot learning testbed with Optimus
Independent research by Iacon: building the demonstration, control, and evaluation workflow for a Franka Panda picking task with Optimus.
independent research · AUG 28, 2026
What would you like to improve?
Tell us about your agents, models, or research workflow. We’ll work through where Arkenos could help and what a useful first project would look like. Write to dev@iaconautonomics.com or book a conversation.
Talk to us