{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "ouyang-et-al-reasoningbank-experience-memory-2025",
    "title": "ReasoningBank: learning reusable strategies from successes and failures",
    "summary": "An agent may repeatedly make the same mistake even when a previous attempt exposed it. ReasoningBank turns self-judged successful and failed trajectories into short strategy records with a title, description and content. Later tasks retrieve relevant records, and new experiences add further lessons. Memory-aware test-time scaling, or MaTTS, compares parallel attempts or uses sequential refinement to obtain richer experience. Ouyang and colleagues test the combined approach on web navigation and software issues. The contribution is the particular strategy-memory and scaling method, not the general idea that failures can teach useful lessons.",
    "boundary": "Success and failure labels come from a model and can be noisy. Reduced interaction steps do not include all memory preparation or token costs. MaTTS spends extra computation, and best-of-several results differ from a single attempt. Some web task success rates remain low despite better step-level metrics. Failure on one attempt does not show that a route is impossible. This is not verified autonomous lifelong learning, an adversarially safe memory service or a separate discovery for each tested model.",
    "area": "Experience-based agent memory",
    "topic": "Artificial intelligence",
    "year": 2025,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "ouyang-et-al-reasoningbank-experience-memory-2025",
    "target": {
      "record_id": "urn:uuid:47032cf9-4659-47bd-a2db-2f505bd496f4",
      "version": 1,
      "content_hash": "sha256:cb8dd523b80e742db23af1c48ecea7c48eda1b61b73afd0865b53a91984e7b99"
    },
    "humans": [
      "Siru Ouyang",
      "Jun Yan",
      "I-Hung Hsu",
      "Yanfei Chen",
      "Ke Jiang",
      "Zifeng Wang",
      "Rujun Han",
      "Long T. Le",
      "Samira Daruki",
      "Xiangru Tang",
      "Vishy Tirumalashetty",
      "George Lee",
      "Mahsan Rofouei",
      "Hangfei Lin",
      "Jiawei Han",
      "Chen-Yu Lee",
      "Tomas Pfister"
    ],
    "authors_complete": true,
    "human_role_note": "Source-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred.",
    "agents": [],
    "related": [],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "cur20261017-arxiv-2509-25140v1",
        "title": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
        "url": "https://arxiv.org/abs/2509.25140v1",
        "date": "2025-09-29"
      }
    ],
    "points": "90000",
    "published_grade": {
      "id": "ouyang-et-al-reasoningbank-experience-memory-2025",
      "entry_id": "ouyang-et-al-reasoningbank-experience-memory-2025",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:47032cf9-4659-47bd-a2db-2f505bd496f4",
        "version": 1,
        "content_hash": "sha256:cb8dd523b80e742db23af1c48ecea7c48eda1b61b73afd0865b53a91984e7b99"
      },
      "what_changed": "An agent may repeatedly make the same mistake even when a previous attempt exposed it. ReasoningBank turns self-judged successful and failed trajectories into short strategy records with a title, description and content. Later tasks retrieve relevant records, and new experiences add further lessons. Memory-aware test-time scaling, or MaTTS, compares parallel attempts or uses sequential refinement to obtain richer experience. Ouyang and colleagues test the combined approach on web navigation and software issues. The contribution is the particular strategy-memory and scaling method, not the general idea that failures can teach useful lessons.",
      "scope": "An agent may repeatedly make the same mistake even when a previous attempt exposed it. ReasoningBank turns self-judged successful and failed trajectories into short strategy records with a title, description and content. Later tasks retrieve relevant records, and new experiences add further lessons. Memory-aware test-time scaling, or MaTTS, compares parallel attempts or uses sequential refinement to obtain richer experience. Ouyang and colleagues test the combined approach on web navigation and software issues. The contribution is the particular strategy-memory and scaling method, not the general idea that failures can teach useful lessons. Success and failure labels come from a model and can be noisy. Reduced interaction steps do not include all memory preparation or token costs. MaTTS spends extra computation, and best-of-several results differ from a single attempt. Some web task success rates remain low despite better step-level metrics. Failure on one attempt does not show that a route is impossible. This is not verified autonomous lifelong learning, an adversarially safe memory service or a separate discovery for each tested model.",
      "status": "PROVISIONAL_EDITORIAL",
      "scientific_correctness_certified": false
    },
    "assessment_disposition": {
      "status": "ASSESSED",
      "label": "Provisionally assessed",
      "reason": "A current grade and separate reference score identify this exact contribution.",
      "target": {
        "record_id": "urn:uuid:47032cf9-4659-47bd-a2db-2f505bd496f4",
        "version": 1,
        "content_hash": "sha256:cb8dd523b80e742db23af1c48ecea7c48eda1b61b73afd0865b53a91984e7b99"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-10-23T00:00:00Z",
    "publisher": {
      "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33",
      "role": "WRS_EDITORIAL_PUBLICATION"
    },
    "record": {
      "contributors": [
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Siru Ouyang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jun Yan",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "I-Hung Hsu",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yanfei Chen",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Ke Jiang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Zifeng Wang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Rujun Han",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Long T. Le",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Samira Daruki",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Xiangru Tang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Vishy Tirumalashetty",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "George Lee",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Mahsan Rofouei",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Hangfei Lin",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jiawei Han",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Chen-Yu Lee",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Tomas Pfister",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
            "url": "https://arxiv.org/abs/2509.25140v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "Success and failure labels come from a model and can be noisy. Reduced interaction steps do not include all memory preparation or token costs. MaTTS spends extra computation, and best-of-several results differ from a single attempt. Some web task success rates remain low despite better step-level metrics. Failure on one attempt does not show that a route is impossible. This is not verified autonomous lifelong learning, an adversarially safe memory service or a separate discovery for each tested model.",
        "domain": "Experience-based agent memory"
      },
      "sources": [
        {
          "citation": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
          "url": "https://arxiv.org/abs/2509.25140v1"
        }
      ],
      "statement": "# ReasoningBank: learning reusable strategies from successes and failures\n\nAn agent may repeatedly make the same mistake even when a previous attempt exposed it. ReasoningBank turns self-judged successful and failed trajectories into short strategy records with a title, description and content. Later tasks retrieve relevant records, and new experiences add further lessons. Memory-aware test-time scaling, or MaTTS, compares parallel attempts or uses sequential refinement to obtain richer experience. Ouyang and colleagues test the combined approach on web navigation and software issues. The contribution is the particular strategy-memory and scaling method, not the general idea that failures can teach useful lessons.\n\n## Scientific scope\n\nSuccess and failure labels come from a model and can be noisy. Reduced interaction steps do not include all memory preparation or token costs. MaTTS spends extra computation, and best-of-several results differ from a single attempt. Some web task success rates remain low despite better step-level metrics. Failure on one attempt does not show that a route is impossible. This is not verified autonomous lifelong learning, an adversarially safe memory service or a separate discovery for each tested model.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nSiru Ouyang; Jun Yan; I-Hung Hsu; Yanfei Chen; Ke Jiang; Zifeng Wang; Rujun Han; Long T. Le; Samira Daruki; Xiangru Tang; Vishy Tirumalashetty; George Lee; Mahsan Rofouei; Hangfei Lin; Jiawei Han; Chen-Yu Lee; Tomas Pfister\n\nSource-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred.\n\n\n\n## Original sources\n\n- [ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory](https://arxiv.org/abs/2509.25140v1)\n",
      "summary": "An agent may repeatedly make the same mistake even when a previous attempt exposed it. ReasoningBank turns self-judged successful and failed trajectories into short strategy records with a title, description and content. Later tasks retrieve relevant records, and new experiences add further lessons. Memory-aware test-time scaling, or MaTTS, compares parallel attempts or uses sequential refinement to obtain richer experience. Ouyang and colleagues test the combined approach on web navigation and software issues. The contribution is the particular strategy-memory and scaling method, not the general idea that failures can teach useful lessons.",
      "title": "ReasoningBank: learning reusable strategies from successes and failures"
    },
    "record_id": "urn:uuid:47032cf9-4659-47bd-a2db-2f505bd496f4",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:47032cf9-4659-47bd-a2db-2f505bd496f4",
      "version": 1,
      "content_hash": "sha256:cb8dd523b80e742db23af1c48ecea7c48eda1b61b73afd0865b53a91984e7b99"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:5129d14ebc3b5b487f63aab0c464bb6fe8bfef87d8210a6ca08c4a6ed7e939ec"
  },
  "target": {
    "record_id": "urn:uuid:47032cf9-4659-47bd-a2db-2f505bd496f4",
    "version": 1,
    "content_hash": "sha256:cb8dd523b80e742db23af1c48ecea7c48eda1b61b73afd0865b53a91984e7b99"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "ouyang-et-al-reasoningbank-experience-memory-2025",
    "entry_id": "ouyang-et-al-reasoningbank-experience-memory-2025",
    "revision": 1,
    "title": "ReasoningBank: learning reusable strategies from successes and failures",
    "context": "Experience-based agent memory",
    "track": "Reusable scientific method, experimental result or historical framework",
    "band": "A",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-23",
    "evidence_cutoff": "2026-10-23",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "An agent may repeatedly make the same mistake even when a previous attempt exposed it. ReasoningBank turns self-judged successful and failed trajectories into short strategy records with a title, description and content. Later tasks retrieve relevant records, and new experiences add further lessons. Memory-aware test-time scaling, or MaTTS, compares parallel attempts or uses sequential refinement to obtain richer experience. Ouyang and colleagues test the combined approach on web navigation and software issues. The contribution is the particular strategy-memory and scaling method, not the general idea that failures can teach useful lessons.",
    "scope": "An agent may repeatedly make the same mistake even when a previous attempt exposed it. ReasoningBank turns self-judged successful and failed trajectories into short strategy records with a title, description and content. Later tasks retrieve relevant records, and new experiences add further lessons. Memory-aware test-time scaling, or MaTTS, compares parallel attempts or uses sequential refinement to obtain richer experience. Ouyang and colleagues test the combined approach on web navigation and software issues. The contribution is the particular strategy-memory and scaling method, not the general idea that failures can teach useful lessons. Success and failure labels come from a model and can be noisy. Reduced interaction steps do not include all memory preparation or token costs. MaTTS spends extra computation, and best-of-several results differ from a single attempt. Some web task success rates remain low despite better step-level metrics. Failure on one attempt does not show that a route is impossible. This is not verified autonomous lifelong learning, an adversarially safe memory service or a separate discovery for each tested model.",
    "evidence_summary": "The linked original works support the scoped statement and contribution attribution. WRS has not independently replayed experiments or reconstructed a full mathematical proof. Success and failure labels come from a model and can be noisy. Reduced interaction steps do not include all memory preparation or token costs. MaTTS spends extra computation, and best-of-several results differ from a single attempt. Some web task success rates remain low despite better step-level metrics. Failure on one attempt does not show that a route is impossible. This is not verified autonomous lifelong learning, an adversarially safe memory service or a separate discovery for each tested model.",
    "references": [
      {
        "title": "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory",
        "url": "https://arxiv.org/abs/2509.25140v1",
        "locator": "Original scoped contribution, statement/method and evidence limits; source-specific attribution retained."
      }
    ],
    "roles": [
      {
        "who": "The named authors in the research record",
        "credit": "SCIENCE",
        "did": "Source-listed human researchers and engineers for the exact method or resource. The evaluated learning system is not an independently attested scientific author. Earlier methods, funding and data-collection roles remain separate. Individual effort shares and unresolved person identities are not inferred."
      }
    ],
    "target": {
      "record_id": "urn:uuid:47032cf9-4659-47bd-a2db-2f505bd496f4",
      "version": 1,
      "content_hash": "sha256:cb8dd523b80e742db23af1c48ecea7c48eda1b61b73afd0865b53a91984e7b99"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "90000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}