{"version":"2026-10-19.boundary.1","asOf":"2026-10-19","month":"2026-10","domain":{"id":"finance","title":"金融 · 资料与专业交付","summary":"本试点聚焦取数、检索、会议和跨文件交付。预测与交易的真实净收益证据另设门槛，不能由文档分数推得。","frontierId":"F20","sourceIds":["BS01"],"terms":[{"term":"财报取数 / 可执行推理","definition":"从表格和文字找到数字、财年和单位，再生成可运行公式；算对答案不等于财务解释全部正确。"},{"term":"Agent框架 / 验收项","definition":"框架提供模型可调用的工具与流程；验收项逐条检查来源、数字、解释等要求。"},{"term":"宏平均 / 多数票","definition":"先计算每道题满足验收项的比例，再给每题同等权重平均；三个模型多数票不是专业人士审定。"},{"term":"CRM / 样本外净收益","definition":"CRM保存客户工作记录；样本外用未参与开发的数据检验，净收益还需扣除交易和运营成本。"}]},"focusTask":null,"caseStudy":null,"caseOptions":[{"id":"F01","title":"财报计算：把取数、运算、对账分别验收","lead":"让AI找数和写公式，让程序计算；你核对财年、单位、口径和原文。计算正确仍可能引用错数据。"}],"tasks":[{"id":"F01","domainId":"finance","title":"从财报取数并计算指标","input":"财报表格/文字、财年与问题","output":"数值、公式和页码依据","human":"核对单位、财年、重述与会计口径","acceptance":["每个输入能定位原文","公式可复算","财年/币种一致"],"relation":"FinQA、FinanceBench的限定公开问题；企业私有报表效果未知","workTaskId":null,"actualCoverage":null},{"id":"F02","domainId":"finance","title":"检索金融资料并整理研究依据","input":"研究问题、允许资料与信息截止","output":"有来源的研究摘要/回答","human":"核对时点、原文、遗漏与解释","acceptance":["每条主张有支持材料","截止日之后信息不得混入","不把条款分当整份报告合格"],"relation":"机构检索流程与公开金融研究基准；不等于自动投资决策","workTaskId":null,"actualCoverage":null},{"id":"F03","domainId":"finance","title":"把客户会议变成记录与跟进草稿","input":"已同意处理的会议内容","output":"纪要、行动项与待发送草稿","human":"取得已有同意、修订并决定发送","acceptance":["关键决定准确","行动负责人与日期核对","客户内容权限符合实际授权"],"relation":"机构公开产品流程；本试点不上传客户材料或测量效率","workTaskId":null,"actualCoverage":null},{"id":"F04","domainId":"finance","title":"生成相互一致的投行交付文件","input":"项目资料、模板与核算假设","output":"Excel、幻灯片和备忘录","human":"审查公式、假设、客户要求与发送","acceptance":["文件间数值一致","表格公式可复算","专业人士审核最终产物"],"relation":"BankerToolBench初级投行公开任务；不是完整金融业务","workTaskId":null,"actualCoverage":null},{"id":"F05","domainId":"finance","title":"把研究结论转成可执行投资决定","input":"实时市场、授权、风险/交易成本约束","output":"可审计决策与真实交易回执","human":"承担组合风险、权限及真实收益核算","acceptance":["前瞻样本外表现","扣除全成本的真实净收益","风险与客户约束全部通过"],"relation":"本次入选材料不足以支持完整委派；空白表示未收录适配证据","workTaskId":null,"actualCoverage":null}],"events":[{"id":"F01-2021","taskId":"F01","availableAt":"2021-09-01","eventDate":null,"dateBasis":"按该条支持材料的公开日期进入时间轴；不是实验统一发生日","title":"数字回答开始附可执行推理程序","kind":"benchmark","mode":"bounded","result":"FinQA共8281个专家标注问题，其中1147题为测试集。FinQANet（RoBERTa-large）检索文字与表格并生成可执行计算程序，测试执行准确率65.05%。","aiRole":"检索财报数字并生成可执行计算程序；指定1147题测试集执行准确率65.05%。","config":"固定首稿的1147题测试集、规则表格布局；BERT-base检索器取排名前3条事实，RoBERTa-large程序生成编码器。","metric":"1147题测试集执行答案准确率；不是8281题全量测量或通用财报分析通过率","meaning":"取数、公式和计算可分别核对；这条路径早于通用生成式Agent。","limit":"不是65%的财务工作自动化；专家参照与模型不是同预算增长序列。","sourceIds":["BF01"],"comparison":"任务/配置或指标不同，不计算统一增长率","supersedes":null,"actualCoverage":null,"measurementWindow":null},{"id":"F01-tools-2023","taskId":"F01","availableAt":"2023-10-23","eventDate":null,"dateBasis":"按所引用固定版本的公开日进入；不是首次发现或运行日","title":"把财报问答的计算交给Python","kind":"benchmark","mode":"assist","result":"FinQA测试1,147题；表2同一Codex底层模型的少样本贪心结果为CoT 40.4%、PoT 64.5%。","aiRole":"模型表达变量/计算程序，Python执行；人仍需核对真实财报的取数与口径。","config":"code-davinci-002；Python3.8/SymPy；按数据集使用4–8示例；比较不同提示/执行系统","metric":"财报问答答案指标；CoT使用相对容差0.001，非完整报告验收","meaning":"可计算步骤有外部执行接口，但正确运算仍可能基于错误取数。","limit":"历史模型和限定问答；不是现场报表流程、审查人时或岗位覆盖。","sourceIds":["BF06"],"comparison":"只解释这一来源/任务/配置；不与其他题集连成增长曲线","supersedes":null,"actualCoverage":null,"measurementWindow":null},{"id":"F01-2023","taskId":"F01","availableAt":"2023-11-20","eventDate":null,"dateBasis":"按该条支持材料的公开日期进入时间轴；不是实验统一发生日","title":"检索方式决定财报问答可靠性","kind":"benchmark","mode":"bounded","result":"FinanceBench公开150题中，GPT-4-Turbo共享向量库29/150答对；整份财报长上下文118/150答对。","aiRole":"模型从财报回答问题；同150题的正确数因资料流程不同为29或118。","config":"同一论文不同配置；长文档超限时截断至约95k–100k tokens","metric":"人工核对的正确回答数；包含逻辑与计算依据检查","meaning":"这组实验中，同一模型因资料处理方式不同表现悬殊；实际使用时仍要核对材料、页码与口径。","limit":"这是同论文、同实验窗口中的配置比较，不是两次时间节点的增长。","sourceIds":["BF02"],"comparison":"同一150题的配置差异可比较；不能当历史增长","supersedes":null,"actualCoverage":null,"measurementWindow":{"start":"2023-11","end":"2023-11","precision":"month","label":"2023年11月测试；Appendix B仅给月份，具体日期未知；不是整月连续测量"}},{"id":"F02-2024","taskId":"F02","availableAt":"2024-06-26","eventDate":"2023-09上线（2024-06公告回顾）","dateBasis":"按该条支持材料的公开日期进入时间轴；不是实验统一发生日","title":"金融机构确认内部研究资料检索上线","kind":"product","mode":"assist","result":"Morgan Stanley公告回顾：Assistant于2023年9月全面推出，为顾问提供内部知识检索。","aiRole":"机构产品帮助顾问查找内部研究资料；此材料没有测得任务净人时。","config":"机构内部资料与顾问核对；本条仅取官方流程说明","metric":"已提供的机构工作流；无本条净人时试验","meaning":"资料查找和摘要可进入员工工具，建议与客户关系继续由人负责。","limit":"证据公告在2024年6月；不把回顾信息倒填成2023已知公开证明。","sourceIds":["BF03"],"comparison":"任务/配置或指标不同，不计算统一增长率","supersedes":null,"actualCoverage":null,"measurementWindow":null},{"id":"F03-2024","taskId":"F03","availableAt":"2024-06-26","eventDate":null,"dateBasis":"按该条支持材料的公开日期进入时间轴；不是实验统一发生日","title":"会议纪要连接到跟进草稿和CRM","kind":"product","mode":"assist","result":"Morgan Stanley Debrief在客户同意后生成会议笔记、行动项与邮件草稿，并保存Salesforce笔记；顾问修订和决定发送。","aiRole":"产品在客户同意后起草纪要、行动项、跟进邮件并记录CRM。","config":"机构工具、OpenAI支持、客户同意与顾问审阅","metric":"官方公开功能流程；没有本试点成功率/实测净人时","meaning":"文稿准备可以减少手工步骤，但客户决定和发送责任保留。","limit":"机构自报，不能把采用率变成纪要准确率。","sourceIds":["BF03"],"comparison":"任务/配置或指标不同，不计算统一增长率","supersedes":null,"actualCoverage":null,"measurementWindow":null},{"id":"F04-2026","taskId":"F04","availableAt":"2026-04-13","eventDate":null,"dateBasis":"按该条支持材料的公开日期进入时间轴；不是实验统一发生日","title":"跨文件交付成为完整任务测试","kind":"benchmark","mode":"bounded","result":"BankerToolBench含100项初级投行任务；21位银行人员对GPT-5.4/Opus4.6产物172次评审中，没有评为可直接发送。","aiRole":"终端Agent生成投行表格、幻灯片与备忘录；指定专业评审中无可直接发送档。","config":"终端Agent编辑Excel、幻灯片与备忘录；人工Readiness评审","metric":"可直接发送档；172是评审次数，不是172个独立完整任务","meaning":"AI可以先产出草稿，但发送前仍需要数字、公式和跨文件审查。","limit":"0%仅适用这次指定评审；不是行业零能力，也不是没有可用草稿。","sourceIds":["BF04"],"comparison":"任务/配置或指标不同，不计算统一增长率","supersedes":null,"actualCoverage":null,"measurementWindow":null},{"id":"F02-2026","taskId":"F02","availableAt":"2026-08-12","eventDate":null,"dateBasis":"按该条支持材料的公开日期进入时间轴；不是实验统一发生日","title":"金融研究开始逐项核对来源与分析","kind":"benchmark","mode":"bounded","result":"FrontierFinance含220查询、11543条附来源的验收项；Samaya高努力内部Agent框架宏平均56.0%，Fable5+Finance Agent v2为49.2%。","aiRole":"系统多轮检索和组合研究材料；最高内部配置的验收项满足比例等权平均56.0%。","config":"Samaya内部框架与Finance Agent v2均使用200工具调用/300秒预算，写入提示，触及后要求最终回答；两套工具/模型配置仍不同。","metric":"GPT5.4、Gemini3.1 Pro、Claude Sonnet4.6三模型多数票判断每个验收项是否满足；先算每题满足比例，再对220题等权平均（宏平均）。不是整题通过率或专业人士批准。","meaning":"Agent能执行多步研究，但数字、时点、比较与结论仍要逐项验收。","limit":"不与FinanceBench正确回答率相连，也不表示56%的投研工作可自动化。","sourceIds":["BF05"],"comparison":"任务/配置或指标不同，不计算统一增长率","supersedes":null,"actualCoverage":null,"measurementWindow":null}],"sources":[{"id":"BF01","title":"FinQA固定首稿","url":"https://arxiv.org/html/2109.00122v1","publishedAt":"2021-09-01","kind":"研究基准","locator":"§3任务；§4.4数据切分；§6设置（BERT-base/top3）与§6.1表2（RoBERTa-large）。","limit":"65.05%为FinQANet执行准确率；91.16%专家参照不作同样本同预算受控增长比较","family":"BF01","readAt":"2026-10-14","readingScope":"本轮直接阅读所列原文位置；未复跑实验","dateBasis":"页面标示发布日期或固定预印本v1页首日期","status":"active"},{"id":"BF02","title":"FinanceBench固定首稿","url":"https://arxiv.org/html/2311.11944v1","publishedAt":"2023-11-20","kind":"研究基准","locator":"§4配置及§4.1标注；§5表2（Shared Vector Store与Long Context）。 Appendix B（测试月份）。","limit":"150题公开子集；29/150与118/150是同论文实验窗口中的不同流程，不能画成时间进步","family":"financebench","readAt":"2026-10-15","readingScope":"2026-10-15直接复核§4/§5及Appendix B；未复跑实验","dateBasis":"页面标示发布日期或固定预印本v1页首日期","status":"active"},{"id":"BF03","title":"Morgan Stanley Debrief正式公告","url":"https://www.morganstanley.com/press-releases/ai-at-morgan-stanley-debrief-launch","publishedAt":"2024-06-26","kind":"机构产品公告","locator":"client consent；Advisor edit/send；Salesforce；Assistant的2023年9月回顾","limit":"机构自报流程/采用，不是独立效率试验；2023上线信息此处到2024才公开确认","family":"morganstanley-ai","readAt":"2026-10-14","readingScope":"本轮直接阅读所列原文位置；未复跑实验","dateBasis":"页面标示发布日期或固定预印本v1页首日期","status":"active"},{"id":"BF04","title":"BankerToolBench固定首稿","url":"https://arxiv.org/html/2604.11304v1","publishedAt":"2026-04-13","kind":"研究基准","locator":"§3任务与环境；§6.1；附录A.4人工评审","limit":"21名银行人员172次评审，指定GPT-5.4/Opus4.6产物；0%可直接发送不是全金融行业结论","family":"BF04","readAt":"2026-10-14","readingScope":"本轮直接阅读所列原文位置；未复跑实验","dateBasis":"页面标示发布日期或固定预印本v1页首日期","status":"active"},{"id":"BF05","title":"FrontierFinance固定首稿","url":"https://arxiv.org/html/2608.11683v1","publishedAt":"2026-08-12","kind":"研究基准","locator":"§3任务与来源化验收项；§5共同预算；§5.1三模型多数票及宏平均；表4系统/框架结果。","limit":"56.0%是Samaya高努力内部harness的条款满足宏平均，不是完整研究通过率","family":"BF05","readAt":"2026-10-14","readingScope":"本轮直接阅读所列原文位置；未复跑实验","dateBasis":"页面标示发布日期或固定预印本v1页首日期","status":"active"},{"id":"BS01","title":"金融NLP系统综述","url":"https://arxiv.org/html/2504.07274v1","publishedAt":"2025-04-09","kind":"综述","locator":"§2筛选；§3任务；§7.1指标与§7.2危机期间","limit":"374候选/221金融任务论文；论文数不是能力分/真实交易净收益","family":"BS01","readAt":"2026-10-14","readingScope":"本轮直接阅读所列原文位置；未复跑实验","dateBasis":"页面标示发布日期或固定预印本v1页首日期","status":"active"},{"id":"BF06","title":"Program of Thoughts：外部计算接口（v4）","url":"https://arxiv.org/html/2211.12588v4","publishedAt":"2023-10-23","kind":"数值任务研究预印本","locator":"§2.2；§3.1数据/实现/指标；§3.2表2","limit":"同底层模型改变提示与执行流程；未匹配完整延迟/预算；不是端到端财务工作通过率","family":"program-of-thoughts","readAt":"2026-10-15","readingScope":"本轮直接核对所列原文位置/版本；未复跑外部实验","dateBasis":"固定v4页首：2023-10-23；不把v4表格倒填v1","status":"active"}],"scope":"16个编辑选择的代表任务切片；不是完整职业清单或任务覆盖比例。","actualCoverage":null,"knowledgeRule":"主轴按原始材料公开日期；实验/上线时点另列。历史空白仅表示本次未收录证据；后来的质疑不偷偷改写早期结果。","method":{"coverage":"覆盖的含义：在指定输入、环境和验收下，材料支持AI参与哪些动作；未知比例保持null。","speed":"速度的含义：证据节点之间的日历间隔、流程交接变化；指标/样本/预算不一致时，不计算能力增长率或外推一年三年。","sourcePolicy":"综述用于发现任务和原始文献；每条公开主张引用原始论文/官方记录。转载及同一文献家族不累计成独立验证。","selection":"目的性选择三个领域的代表任务，不是穷尽检索或无偏职业采样；基础时间轴于2026-10-14核对，本次精选补充与证据截止2026-10-15，固定论文优先，不把动态榜排名倒填历史。"}}