起步顺序Order of work 先证明内部收益,再开一个有人兜底的客户渠道,最后扩展到资本市场与美洲业务。Prove the internal gains, then open one customer channel with a person behind it, then widen to markets and the Americas.
公开事实Public facts 已核对Checked
- 71,000+截至 2026 年 6 月,可用辅助 AI 工具的员工Employees with assistive AI tools by June 2026基于 Scotia Intelligence 平台的 Scotia NavigatorScotia Navigator, on the Scotia Intelligence platform
- 5,500用 AI 编程的工程师Engineers coding with AI公司披露Disclosed by Scotiabank
- 70%商业邮件人工工作量的降幅Manual workload cut on business emails约 90% 的此类邮件由 AI 处理About 90% of those emails are handled by AI
- 30thEvident 2026 AI 指数,50 家银行中排名Evident 2026 AI Index rank among 50 banks尽管内部铺开很广Despite the broad rollout
从员工到客户的四级台阶Four steps from staff to customers 判断Judgment
- 员工助手Staff assistants:已上线,7.1 万+员工Live, 71,000+ employees
- 客服,有人兜底Customer service, person behind:在客服中心基础上建Build on the contact centre
- 客户理财引导Customer guidance:须先过行为合规评审After a conduct review
- 资本市场与美洲业务Markets and the Americas:文件、多语言、信用Documents, languages, credit
专家组裁决Expert panel 依据公开观点的推演Inference from public views
Foteini Agrafioti
把 Scotia Intelligence 挂到客户目标上;平台与业务条线的权责写成一页纸。Tie the platform to customer targets; one page on who owns what.
Erik Brynjolfsson
先给 40% 与 70% 的成果定基线;预算含流程再造与培训,并优先增强员工。Baseline the 40% and 70% claims; budget for process redesign.
Ziad Obermeyer
信贷评估或理财建议上线前,先写清真实目标与代理指标;按客户群分别看结果。For advice and credit, name the real goal and the proxy first.
Ethan Mollick
开放客户渠道前,先在真实通话中测出助手的失败点;保留人在回路中。Test where the assistant fails on real calls; keep a person in the loop.
Andrew Ng
选一个三到六个月内出数字的客户任务,并写明成功后复制到哪。Pick one customer task that shows numbers in three to six months.
Thomas Davenport
50 家中排第 30 暗示广而不深;点名哪些条线的流程真正改变。A 30th-of-50 rank hints at breadth, not depth; name the lines that change.
Ethan Mollick:尽早让客户接触助手,有人兜底。Let customers meet the assistant early.
Ziad Obermeyer:理财建议与信贷模型先过目标检查。Advice and credit pass a target check first.
→ 客服任务走快速通道;理财建议与信贷走完整审查。Fast lane for service tasks; full review for advice and credit.
六位专家是谁、怎样思考、有什么铁律 →Who the six experts are, how they think, their rules →
三个目标Three priorities 建议Recommendation
给已披露的成果定基线Baseline the gains already claimed
把 40% 与 70% 变成按业务条线可核对的数字。Turn 40% and 70% into audited numbers by business line.
开一个客户渠道Open one customer channel
把客服中心扩展到办完一个简单请求,由人兜底。Extend the contact centre to resolve a simple request, with a person behind.
向其他业务复用Reuse across markets
把已验证的能力复用到国际银行与资本市场。Carry proven skills to international banking and capital markets.
场景组合Scenarios 建议Recommendation
| 场景Scenario | 做法Approach | 状态Status |
|---|---|---|
| 客服端到端办结Customer service, resolved end to end 体验Experience提效Efficiency | 从答复咨询升级为办结简单请求,由人确认。Move from answering enquiries to resolving a simple request, with a person confirming. | 建议 · 基于已披露的 40%Proposed · builds on the 40% disclosed |
| 个性化理财辅助Personalised guidance support 增收Revenue降风险Risk | 在行为合规评审后,为顾问与客户提供个性化建议。Help advisers and customers with tailored suggestions after a conduct review. | 建议 · 完整审查Proposed · full review |
| 美洲客服与信用Americas service and credit 体验Experience降风险Risk | 在国际银行业务中做多语言客服与信用评估。Multilingual service and credit assessment across its international banking footprint. | 建议Proposed |
| 交易文件比对Trade document comparison 提效Efficiency降风险Risk | 延伸其已获批的 AI 辅助确认文件比对专利,并加研报摘要。Extend its granted patent on A I-assisted confirmation comparison, plus research summaries. | 建议 · 专利 2025 年 3 月获批Proposed · patent granted March 2025 |
12 个月路线12-month roadmap 建议Recommendation
奠基Foundation
- 40% 与 70% 的基线Baselines for the 40% and 70%
- 模型清单与风险分级Model inventory and risk tiers
- 一页权责One page of ownership
试点Pilot
- 一个客户任务上线One customer task live
- 真实通话测试Test on real calls
- 对照组测量Control group measured
上线Go live
- 任务向全体客户开放Task opens to all customers
- 关键步骤人工确认Person confirms key steps
- 上线即监控Monitoring from day one
扩展Extend
- 理财建议准入评审Guidance gate review
- 美洲业务复用Reuse in the Americas
- 年度重排优先级Annual re-ranking
治理与合规Governance 建议Recommendation
- 模型清单对齐 OSFI E-23(2027-05-01 生效)Model inventory aligned to OSFI E-23 (effective 2027-05-01)
- 理财建议与信贷模型:完整审查Advice and credit models: full review
- 涉及客户权益的步骤由人确认A person confirms customer-affecting steps
三层指标Three-layer metrics 建议Recommendation
| 采用Adoption | 各条线周活跃使用率Weekly active use by line · 接触助手的客户数Customers reaching the assistant |
|---|---|
| 过程Process | 无需人工的办结率Resolution rate without a person · 处理时长Handling time · 转人工率Escalation rate to a person |
| 结果Outcome | 单个已办结请求成本Cost per resolved request · 客户满意度对照基线Customer satisfaction against baseline |
目标值由企业自定。Targets are the company's to set.
风险Risks 判断Judgment
内部成果无法直接迁移Internal wins do not transfer
员工工具容得下错误,客户容不下。先在真实通话中测试。Staff tools forgive errors; customers do not. Test on real calls first.
成果缺乏基线Claims without baselines
没有基线的百分比,难以向董事会或监管机构交代。Percentages without baselines are hard to defend to a board or a regulator.
理财与信贷中的代理指标Proxy goals in advice and credit
模型可能优化销量而非客户需求;上线前写清两者。A model can optimise sales, not customer need; name both before launch.
前 90 天First 90 days 建议Recommendation
| 第 1–2 周Weeks 1–2 | 点名负责人;定基线Name owners; fix baselines |
|---|---|
| 第 3–4 周Weeks 3–4 | 盘点所有在用模型与助手Inventory every live model and assistant |
| 第 5–10 周Weeks 5–10 | 风险分级初稿;选定一个客户任务Risk tiers drafted; pick one customer task |
| 第 13 周Week 13 | 首次复盘First review |
我们可能错在:We could be wrong if: 客户在涉钱问题上仍坚持找真人,或 40% 只统计了本来就简单的咨询。customers still ask for a person on money matters, or the 40% counts only enquiries that were already easy.
本页是基于公开资料的分析,与所述企业没有合作关系,不代表其官方立场或内部计划;企业数字多来自其自述或媒体转述;「专家组」是依据六位专家公开观点的推演,不是他们的原话,也不代表他们或其雇主的立场。资料整理日期:2026-11-14。This page is an analysis based on public information. It has no affiliation with the company and does not represent its official position or internal plans. Company figures come mostly from its own statements or media reports. The expert panel is an inference from six experts' public views, not their words, and does not represent them or their employers. Compiled 2026-11-14.
想给你的公司也做一份?Want one for your company?
同样的方法:公开事实、专属判断、12 个月路线和前 90 天。公司大小都可以聊,中文沟通。Same method: public facts, a company-specific judgment, a 12-month roadmap and the first 90 days. Any size of business; we can talk in Chinese or English.