Codos reports Aethos X scores 94% on EnterpriseRAG-Bench

Source published 15m agoTerminal published 13m ago

Summary

Codos, a virtual Chief AI Officer, is addressing the issue that Forward Deployed Engineers face with knowledge retention, as their understanding of company systems often fades after projects end. By interviewing employees and deploying a strategy to create a company-wide memory layer, Codos has reportedly helped a fintech client free up 21% of its FTE capacity in just six months, contributing over $10 million in impact. Codos' Aethos X system achieved a preliminary score of 94% on the EnterpriseRAG-Bench, outperforming parallel GPT-6 Astra agents. This benchmark shows significant progress in structured memory systems, which are crucial for maintaining the complex, scattered records typical in enterprise environments and essential for developing self-improving AI agents.

Analysis

Codos: Codos develops the first virtual Chief AI Officer platform designed to help companies capture and retain institutional knowledge that is typically lost at the end of consulting engagements. The system interviews employees to identify automation opportunities, deploys a company-wide memory layer with agents, and executes transformation projects, including through on-premises installations. In this news, Codos is positioned as the solution addressing the compounding knowledge-loss problem described in the announcement and blog post. Aethos: Aethos is the graph-based memory system that powers the Codos Company Brain by extracting facts and relationships from documents and conversations to create connected, maintainable knowledge records. It organizes information prior to queries, enabling agents to follow relationships back to source evidence and preserve historical context as new data arrives. In this news, Aethos X is the experimental configuration benchmarked against other systems on large enterprise-style corpora, demonstrating architectural advantages in retrieval quality. Dima Khanarin: Dima Khanarin is the voice and key figure introducing Codos as a virtual Chief AI Officer that converts AI capabilities into measurable business impact for enterprises. He highlights the platform's ability to run on customer servers while delivering ongoing value to NASDAQ-listed and PE-backed clients. In the news, he is quoted directly on the product launch and shares preliminary benchmark results for its underlying Aethos technology. AI Memory Progress: Advances in graph-based memory architectures are viewed as foundational for enabling self-improving agents that retain context from prior work and feedback. Knowledge Retention: Enterprise environments generate complex, scattered records such as emails, Slack messages, tickets, and meeting transcripts that require structured memory systems to make evidence discoverable for agents. Enterprise AI Impact: Despite strong performance on benchmarks, many companies continue to face difficulties translating AI capabilities into direct profit-and-loss results.

Summary

Analysis

Categories