Research
Explore our ongoing and completed research initiatives across key AI domains.
Autonomous Coding Agents
Investigating headless, non-interactive parallel execution of coding agents for large-scale software development, focusing on reasoning, code quality, throughput, and LLM flexibility.
Multi-Agent System Coordination
Explore open protocols for AI interoperability (A2A, MCP, OVON) to enable seamless collaboration across agents.
Key Findings:
- Interoperability requires standardized protocols for inter-agent communication (A2A, OVON) and agent data/tool access (MCP).
- A2A provides a framework for general agent coordination, discovery (Agent Cards), and secure communication.
- OVON focuses on interoperability using a 'Conversation Envelope' and dialogue events for managing handoffs and context.
- These protocols aim to enable collaboration between specialized agents, breaking down vendor/framework silos.
- Adoption hurdles, security, and standardization competition are key challenges for these emerging coordination frameworks.