--- schema: formation.pattern_findings/v0.1 kind: pattern_findings visibility: public canonical_url: https://topologyindex.com/patterns/architecture_search/findings.md description: 'What published studies found about architecture_search, in words: each finding''s sentence, comparison, domain, caveat and source. Attributed, stated without figures, never a ranking.' path: /patterns/architecture_search/findings.md pattern: architecture_search pattern_index: /patterns/index.md pattern_page: /patterns/architecture_search.md product_api_version: v1 published_findings: - benchmarks: - DROP - MGSM compared_against: 'State-of-the-art hand-designed agents' direction: helped source_id: arxiv:2408.08435 task_domain: 'Reading comprehension, math, science and coding' task_domains: - coding - reasoning url: https://arxiv.org/abs/2408.08435 - benchmarks: [] compared_against: 'Human-designed agents, accounting for design plus deployment cost' direction: hurt source_id: arxiv:2510.06711 task_domain: 'Automated agent design across several datasets' task_domains: [] url: https://arxiv.org/abs/2510.06711 - benchmarks: [] compared_against: 'Manually designed workflows and prior automated methods' direction: helped source_id: arxiv:2410.10762 task_domain: 'Question answering, code generation and math' task_domains: - coding - reasoning url: https://arxiv.org/abs/2410.10762 - benchmarks: [] compared_against: 'Handcrafted and automated multi-agent systems' direction: helped source_id: arxiv:2502.04180 task_domain: 'Math, coding and tool use' task_domains: - coding - reasoning - operations url: https://arxiv.org/abs/2502.04180 - benchmarks: - MATH - DROP - HotpotQA - MuSiQue - '2WikiMQA' - MBPP - HumanEval - LiveCodeBench compared_against: 'Unoptimized topologies and agent-scaling strategies such as self-consistency and debate' direction: mixed source_id: arxiv:2502.02533 task_domain: 'Reasoning, multi-hop question answering and coding' task_domains: - coding - reasoning url: https://arxiv.org/abs/2502.02533 schema_version: v0.1 title: 'Published findings on the architecture_search multi-agent pattern' --- # Published findings on the architecture_search multi-agent pattern The findings filed under the [architecture_search](/patterns/architecture_search.md) pattern page, in words. That page has the definition, failure modes and sources; the [pattern catalogue](/patterns/index.md) has the general findings about multi-agent systems. ## What published studies found Attributed to each source and stated without figures, because a number from one configuration reads as a result for the pattern. Unfavourable results are included on purpose. None of this is evidence produced by this service. Reviewed 2026-09-23. Each label says how this pattern fared against what it was compared with, as the source reports it; the labels are defined at [/docs/schemas/pattern/v0.1.md](/docs/schemas/pattern/v0.1.md). - **Helped** — ADAS reports that agents discovered by a meta agent programming new designs in code outperformed hand-designed agents and kept their advantage when transferred across domains and models. Compared against: State-of-the-art hand-designed agents. Domain: Reading comprehension, math, science and coding. Benchmarks: DROP, MGSM. Caveat: Search cost is not weighed against the gains in the abstract, and later work questions its cost-effectiveness. Source: [Automated Design of Agentic Systems](https://arxiv.org/abs/2408.08435), Hu, Lu, Clune, 2024-08-15. - **Hurt** — El and colleagues report that meta-agent design pays for itself against human-designed agents only on a small minority of datasets, and that feeding all prior designs into context performs worse than ignoring them. Compared against: Human-designed agents, accounting for design plus deployment cost. Domain: Automated agent design across several datasets. Caveat: Analyzes a particular family of meta-agent methods; conclusions depend on deployment volume assumptions. Source: [Inefficiencies of Meta Agents for Agent Design](https://arxiv.org/abs/2510.06711), El, Yuksekgonul, Zou, 2025-10-08. - **Helped** — AFlow reports that Monte Carlo tree search over code-represented workflows beat prior baselines and let smaller models outperform a much larger model on some tasks at a fraction of its inference cost. Compared against: Manually designed workflows and prior automated methods. Domain: Question answering, code generation and math. Caveat: Author-run evaluation; the cheaper-model result holds only on specific tasks. Source: [AFlow: Automating Agentic Workflow Generation](https://arxiv.org/abs/2410.10762), Zhang et al., 2024-10-14. - **Helped** — MaAS reports that sampling query-dependent agent architectures from a learned supernet matched or beat existing systems while using only a fraction of their inference cost, with cross-dataset and cross-model transfer. Compared against: Handcrafted and automated multi-agent systems. Domain: Math, coding and tool use. Caveat: Author-run comparison; some reported accuracy gains are small. Source: [Multi-agent Architecture Search via Agentic Supernet](https://arxiv.org/abs/2502.04180), Zhang et al., 2025-02-06. - **Mixed** — MASS reports that only a small fraction of topologies improved performance, with most failing to help or degrading it, and that prompt optimization was more token-effective than scaling agent count. Compared against: Unoptimized topologies and agent-scaling strategies such as self-consistency and debate. Domain: Reasoning, multi-hop question answering and coding. Benchmarks: MATH, DROP, HotpotQA, MuSiQue, 2WikiMQA, MBPP, HumanEval, LiveCodeBench. Caveat: Findings come from one search framework and a limited set of model backbones. Source: [Multi-Agent Design: Optimizing Agents with Better Prompts and Topologies](https://arxiv.org/abs/2502.02533), Zhou et al., 2025-02-04. ## How evidence works here A pattern names how work is organized, never how well that works. Evidence in this service attaches to a complete execution configuration — a topology formation together with an execution lock pinning models, tools and runtime — and never to a pattern on its own, so no page here carries a success rate, a cost or a run count for an arrangement.