hiloop provides infrastructure designed to support recursive self-improvement for AI systems. The platform assists teams in training agents for tasks where general models fall short. Users supply a task, their current agent or model, and an evaluation framework. hiloop then conducts an automated research campaign spanning data, supervised fine-tuning, other post-training methods, continual learning, prompts, tools, harnesses, and systems. The result is the best verified improvement, delivered either through hiloop's hosted environment or within the user's own cloud infrastructure.
The company offers a research system around models, incorporating persistent memory, complete experiment lineage, compute orchestration, and statistical verification. Initial focus areas include agent and model training, continual learning, and optimization. Teams interested in early access can contact hiloop directly via email. The service is listed on Nova Startup, a modern premium directory for global startups, and falls under the artificial intelligence category with tags including developer tools, machine learning, reinforcement learning, and infrastructure.