Company AI fingerprint

Two companies, six capability axes, one chart. Where each polygon bulges tells you what that company is actually building. The asymmetries — not the absolute values — are the story.

Company A:
Company B:
Foundation trainingFoundation training OpenAI: 27% (61/228) Capital One: 55% (120/219)27% / 55%Inference infraInference infra OpenAI: 48% (109/228) Capital One: 64% (141/219)48% / 64%Agent platformAgent platform OpenAI: 36% (83/228) Capital One: 33% (72/219)36% / 33%Eval & safetyEval & safety OpenAI: 58% (132/228) Capital One: 52% (113/219)58% / 52%RAG / retrievalRAG / retrieval OpenAI: 13% (30/228) Capital One: 59% (129/219)13% / 59%Robotics + multimodalRobotics + multimodal OpenAI: 7% (17/228) Capital One: 5% (11/219)7% / 5%Research depthResearch depth OpenAI: 14% (32/228) Capital One: 19% (42/219)14% / 19%OpenAI · Foundation training: 27% (61/228)OpenAI · Inference infra: 48% (109/228)OpenAI · Agent platform: 36% (83/228)OpenAI · Eval & safety: 58% (132/228)OpenAI · RAG / retrieval: 13% (30/228)OpenAI · Robotics + multimodal: 7% (17/228)OpenAI · Research depth: 14% (32/228)Capital One · Foundation training: 55% (120/219)Capital One · Inference infra: 64% (141/219)Capital One · Agent platform: 33% (72/219)Capital One · Eval & safety: 52% (113/219)Capital One · RAG / retrieval: 59% (129/219)Capital One · Robotics + multimodal: 5% (11/219)Capital One · Research depth: 19% (42/219)OpenAI· 228 AI roles, 9 regulatedCapital One· 219 AI roles, 13 regulated
Each axis = share of company’s tagged AI roles touching that capability cluster. Outer ring = 100%, inner ring = 25%. Hover any vertex for the underlying count.

Suggested comparisons