Apodex-1.0-H has launched an advanced agent team designed for deep research, achieving state-of-the-art results by splitting web research across various agents while meticulously auditing every evidence chain before forming answers. This innovative system treats deep research as a distributed systems challenge, employing an async agent team where an orchestrator assigns distinct contexts and tools to sub-agents. Notably, Apodex showcases a novel approach to inference-time scaling in AI research, emphasizing the coordination of various search agents and a separate verification layer, rather than merely increasing model size, to produce more accurate responses.

Apodex: Apodex is an AI project developing advanced multi-agent systems for complex research tasks. It recently released Apodex-1.0-H, a heavy-duty agent team that splits web research across specialized sub-agents while using dedicated fact-checker, conflict-reviewer, and draft-reviewer agents to audit evidence chains. The system frames deep research as a distributed systems challenge solved through orchestration, persistent traces, and a verification layer.
Apodex AI: Apodex AI is the project behind the Apodex agent platform and its 1.0-H release for coordinated AI research. It emphasizes inference-time scaling via many search agents and separate verification rather than relying on a single larger model. The project shares updates and resources through its official channels including blog posts and technical reports.

Scaling Approach: Apodex demonstrates inference-time scaling by achieving stronger results through coordinated agents, persistent traces, and verification layers instead of model size increases alone.
Agent Coordination: Apodex uses an async agent team where an orchestrator assigns separate contexts and tools to sub-agents before verification steps occur.
Research Methodology: The system treats deep research as a distributed systems problem solved through evidence auditing and conflict review prior to final answer generation.