The economic value of agents will soon come from running whole businesses, not from finishing tasks.
AI & ML interests
Agent evaluation, simulation environments, long-horizon decision-making, and autonomous businesses
Recent Activity
Organization Card
The length of tasks that AI is completing is doubling every 4 months. The rate of change in the development of AI is increasing. The economy is not ready for the autonomy that frontier model capabilities will demonstrate. In order for AI to be a useful, productive economic actor, it needs to learn the tacit knowledge dispersed across many individuals in the most challenging industries.
ORO builds high-fidelity simulation environments grounded in real data to teach AI the tacit knowledge required to run safe and profitable autonomous businesses.
SMB Bench v1.0 measures AI's ability to learn from delayed consequences. Models are tasked with running a produce buying desk for a wholesaler over a 90-day quarter: keeping customers supplied profitably while demand, suppliers, freight and cash all move, in a partially observable environment. No model reliably beats the fixed rule in each run, and only GPT-6 Astra finishes ahead of it on average. Results, traces, the gated grader and the report's figures: https://huggingface.co/collections/oro-agents/smb-bench-6aec4d315223a590f0bc941a.
If it excites you too, we would love to talk: team@oroagents.com.
models 0
None public yet