---
id: "claim-diversity-improves-performance"
type: "claim"
source_timestamps: ["¶4"]
tags: ["performance-metrics", "team-diversity", "software-engineering"]
related: ["claim-two-diverse-beats-sixteen", "concept-cognitive-friction", "concept-structural-ai-diversity"]
confidence: "high"
testable: true
speakers: ["Mark Purdy"]
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-new-28-agent-teams-different-models"
sourceUrl: "https://hbr.org/2026/06/the-strongest-teams-of-ai-agents-will-be-built-using-different-models"
sourceTitle: "The Strongest Teams of AI Agents Will Be Built Using Different Models"
---
# Diverse agent teams significantly outperform individual or homogeneous agents

**Claim:** Agentic diversity yields measurable performance dividends, mirroring the benefits seen in human workforces. One cited study found that agent teams **selected for diversity were 25% better at resolving software-engineering problems** compared to agents acting individually. The improvement is attributed to blending different skills, knowledge sets, and problem-solving approaches — i.e., productive [[concept-cognitive-friction]].

This is the performance case *for* [[concept-structural-ai-diversity]], and it is reinforced by the even more striking efficiency result in [[claim-two-diverse-beats-sixteen]].

**Confidence: high** (as stated) — but see caveat.

**Enrichment assessment:** The *general* claim (heterogeneous multi-agent systems and ensemble methods improve robustness and solution rates) is consistent with active research on multi-agent LLM collaboration and ensemble LLMs for code generation (AWS, IBM, Galileo, ML Mastery all support the idea). The **precise 25% figure** does not match any widely cited, named study in the open literature and appears proprietary or not-yet-widely-referenced. Treat the direction as sound and the exact delta as an **illustrative experimental result**, not a generalizable benchmark.