WorkBuddy × CodeBuddy
Research cut-off · 2026-09-03
商业化场景研究Commercial scenario research

WorkBuddy 与 CodeBuddy
最容易卖在哪里？

结论不是“功能最多的地方”，而是客户愿意付费解决的高频痛点：WorkBuddy 赢在跨系统、跨文档的业务执行；CodeBuddy 赢在真实代码库里的研发提效与治理。

口径说明：腾讯未公开按场景拆分的合同额、订单数或 ARR。本报告的“售卖最多”是基于截至 2026-09-03 可获得的客户数量、案例频率、用户行为、部署规模与量化 ROI 进行的证据化判断，不等同于审计后的收入排名。

Where WorkBuddy and CodeBuddy sell best

The winning scenarios are not those with the longest feature list, but those attached to urgent, repeatable and measurable work: cross-system business execution for WorkBuddy, and repository-level engineering productivity and governance for CodeBuddy.

Scope note: Tencent does not publicly disclose contract value, order count or ARR by scenario. “Sell best” here is an evidence-based inference from disclosed customer counts, case frequency, user behavior, deployment scale and measured ROI as of 3 September 2026—not an audited revenue ranking.

一句话结论

Executive verdict

真正的成交单元不是一个“AI 功能”，而是一条可量化、可复制、可治理的工作流。

The real unit of sale is not an AI feature; it is a measurable, repeatable and governable workflow.

WorkBuddy

卖得最顺：业务数据与材料的“数字员工”

最强横向场景是数据处理、对账、分析、报表和材料生成；最强垂直行业信号是金融。客户通常先买一个高频流程的提效，再扩展到知识检索、合规审核、系统操作和跨部门 Skill 复用。

典型买方CIO / 数字化负责人 / HR / 财务 / 运营 / 业务部门负责人
大单驱动VPC 或私有化、数据不出域、权限与审计、企业微信入口、行业专家与连接器。

Best-selling motion: business-data and document “digital employees”

The strongest horizontal cluster is data processing, reconciliation, analysis, reporting and document production; the clearest vertical scale signal is financial services. Customers land on one frequent workflow, then expand into knowledge retrieval, compliance review, system actions and reusable Skills.

Typical buyerCIO, digital transformation lead, HR, finance, operations and business-function heads.
Large-deal driverVPC/private deployment, data residency, permissions and audit, WeCom entry points, domain experts and connectors.
CodeBuddy

卖得最顺：企业研发团队的存量工程提效

最强场景不是“从零写一个 Demo”，而是理解真实代码库、生成和修改多文件代码、定位 Bug 并补测试。大型客户进一步采购私有化部署、研发系统集成、Rules/知识库与运营度量。

典型买方CTO / 研发效能负责人 / 软件工程中心 / 安全合规 / 开发平台团队
大单驱动开发者席位规模、私有代码安全、遗留系统维护、企业研发流程和 CI/CD 集成。

Best-selling motion: productivity for real enterprise codebases

The strongest scenario is not merely building a demo from scratch, but understanding existing repositories, generating and changing multi-file code, locating defects and adding tests. Larger buyers then add private deployment, R&D-system integration, Rules/knowledge and adoption analytics.

Typical buyerCTO, engineering productivity leader, software center, security/compliance and developer-platform team.
Large-deal driverDeveloper-seat scale, private-code security, legacy maintenance, enterprise workflows and CI/CD integration.

场景优先级

Scenario priority

条形长度代表综合商业信号强度，不是公开收入占比。

Bar length indicates combined commercial evidence, not disclosed revenue share.

WorkBuddy：高频购买场景WorkBuddy: leading purchase scenarios

01
数据处理、对账、分析与报表Data processing, reconciliation, analytics and reporting跨 Excel、业务系统和文档，把数天压缩到分钟级。Compress multi-system and spreadsheet work from days to minutes.
02
金融专业工作流Financial professional workflows信贷尽调、投研投顾、保险经营、合规与材料编制。Credit due diligence, investment research, insurance sales, compliance and documents.
03
知识检索、制度问答与合同审核Knowledge retrieval, policy Q&A and contract review有原文溯源、权限和审计要求的大型组织。Large organizations requiring source traceability, access control and audit.
04
岗位自动化与数字员工Role automation and digital employees把个人经验封装成 Skill，跨部门复制。Package individual know-how into reusable Skills across departments.
05
营销、研究与内容生产Marketing, research and content production行业简报、营销材料、HTML/PPT 与多媒体内容。Industry briefs, sales collateral, HTML/PPT and multimedia assets.
判断：横向使用频率最高的是“数据和材料”；按披露客户数量看，金融是当前最强垂直行业。
Assessment: “data and documents” lead horizontally; finance is the strongest disclosed vertical by customer count.

CodeBuddy：高频购买场景CodeBuddy: leading purchase scenarios

01
工程理解与遗留代码维护Repository understanding and legacy maintenance理解大型存量代码、架构、依赖和历史逻辑。Understand large repositories, architecture, dependencies and legacy logic.
02
代码生成与需求交付Code generation and feature delivery从 Spec 到多文件开发、修改和部署。Move from reviewed specs to multi-file implementation and deployment.
03
Debug、修复与测试Debugging, repair and testing日志定位、缺陷修复、单测生成和代码审查。Log diagnosis, defect repair, unit-test generation and code review.
04
企业级私有部署与研发治理Private deployment and engineering governance代码不出域、Rules、知识库、权限、审计和指标运营。Code residency, Rules, knowledge, permissions, audit and adoption metrics.
05
0→1 应用与 MVPZero-to-one applications and MVPs产品/设计/全栈团队从 PRD、Figma 到可分享应用。From PRD and Figma to a shareable app for product/design/full-stack teams.
判断：0→1 场景利于获客和试用；真正的大型企业采购更偏“存量工程 + 安全治理 + 规模化席位”。
Assessment: zero-to-one work drives trial and acquisition; larger enterprise deals skew toward existing codebases, governance and seat scale.

关键证据

Key evidence

优先采用腾讯官方/开发者社区材料和主流媒体报道；单个案例数据不外推为全市场平均值。

Tencent and mainstream-media evidence is prioritized; single-customer outcomes are not generalized as market averages.

WorkBuddy · Finance
100+

金融机构形成最大披露集群

Largest disclosed vertical cluster

金融版披露已进入中金公司、国投证券、平安银行、中国太平等 100 多家金融机构，覆盖信贷尽调、投研投顾、个客经营与财富资配。

The Financial Edition was reported across 100+ institutions, spanning credit due diligence, investment advisory, customer management and wealth workflows.

Source 01
WorkBuddy · Usage mix
67 / 63 / 57%

垂直业务、数据和自动化并行

Vertical work, data and automation converge

上海家化披露：67% 用户用于垂直业务，63% 用于数据处理与分析，57% 用于自动化与运维；九部门平均提效 4.5 倍以上。

Shanghai Jahwa reported 67% vertical-business usage, 63% data processing/analysis and 57% automation/operations, with 4.5x+ average uplift across nine departments.

Source 02
WorkBuddy · Process
30 → 5 min

规则明确、信息密集的流程最好卖

Structured, information-heavy processes sell well

深圳港报关填报从 30 分钟缩至 5 分钟，并延伸至经营分析、合同审核、招聘筛选和制度查询；数据全程不出企业边界。

Shenzhen Port reduced customs-form preparation from 30 to 5 minutes and expanded into financial analysis, contract review, hiring and policy lookup within enterprise boundaries.

Source 03
CodeBuddy · Usage signal
Top 3

工程理解、生成、Debug

Understand, generate, debug

基于 2,900+ 腾讯内部用户反馈的公开文章，将工程理解、代码生成、Debug 定位与修复列为前三大场景。

A published analysis of feedback from 2,900+ Tencent users identifies engineering understanding, code generation, and debugging/repair as the top three scenarios.

Source 04
CodeBuddy · Enterprise ROI
70%+

存量工程产生最硬 ROI

Existing codebases produce the hardest ROI

TCL 披露编码研发效率提升 70%+；历史代码改造工时缩短 80%+，跨端迁移从 2 周缩至 2 天内，Bug 修复效率提升 80%+。

TCL reported 70%+ coding efficiency uplift, 80%+ less legacy-code effort, cross-platform migration from two weeks to under two days, and 80%+ faster defect repair.

Source 05
CodeBuddy · Scale
8,000+

私有化与研发体系融合带来大规模部署

Private integration enables large-scale rollout

招商银行案例披露内部用户超过 8,000，补全采纳率 28%，AI 代码生成占比 35%+；方案重点是模型、插件、运营闭环。

China Merchants Bank disclosed 8,000+ internal users, 28% completion acceptance and 35%+ AI-generated code, built around model, plugin and adoption-operation loops.

Source 06

最有效的销售路径

Most effective sales motion

两个产品都不是靠“一次性演示”成交，而是靠试点、量化、固化和扩张。

Neither product wins through a one-off demo; both depend on pilot, measurement, codification and expansion.

WorkBuddy

1
选高频、高痛流程Choose a frequent, painful workflow对账、报告、审核、查询、尽调等，最好能拿到现状耗时。Reconciliation, reports, reviews, lookup or due diligence—with a baseline time.
2
小规模种子用户试点Pilot with seed users由业务人员定义规则，AI 执行，人负责复核。Business users define rules; AI executes; humans review.
3
沉淀为 Skill 和数字员工Codify into Skills and digital employees把个人经验变成组织资产，形成权限、审计和可复用流程。Turn personal know-how into governed, reusable organizational assets.
4
跨部门和行业工作台扩张Expand across departments and workbenches从单点席位扩大到企业版、VPC 或私有化。Move from point seats to enterprise, VPC or private deployment.

CodeBuddy

1
选择真实工程和高痛团队Choose a real repository and pain-heavy team遗留代码、迁移、Bug、重复编码比“玩具 Demo”更容易证明价值。Legacy code, migration, defects and repetitive coding prove value better than toy demos.
2
建立基线指标Establish baseline metrics生成率、采纳率、编码时长、缺陷率、测试覆盖和活跃度。Generation, acceptance, coding time, defects, coverage and active use.
3
接入 Rules、知识库和研发系统Integrate Rules, knowledge and R&D systems让工具理解企业代码，并进入需求、评审、测试和 CI/CD。Make the tool understand enterprise code and join requirements, review, testing and CI/CD.
4
从插件扩展为企业 AISEExpand from plugin to enterprise AISE增加私有部署、模型评测、运营推广和度量治理。Add private deployment, model evaluation, adoption operations and governance.
组合销售建议：WorkBuddy 管业务工作流，CodeBuddy 管软件交付工作流Bundle recommendation: WorkBuddy for business workflows, CodeBuddy for software-delivery workflows

同一大型客户中，两者可共享企业知识、身份权限、安全审计、MCP/连接器和 Skills 生态，但应分别向业务负责人和研发负责人证明 ROI。

Within one enterprise, the products can share knowledge, identity, security, connectors and Skills, but ROI should be proven separately to business and engineering buyers.

销售话术矩阵

Sales-message matrix

先讲结果，再讲产品；先讲流程，再讲模型。

Lead with outcomes, not product; lead with workflow, not model.

客户信号WorkBuddy 话术CodeBuddy 话术Customer signalWorkBuddy messageCodeBuddy message
大量 Excel、报告、材料把跨系统找数、整理、校验和成文变成一个可重复执行的数字员工。若背后需要开发自动化系统，用 CodeBuddy 缩短实现与维护周期。Heavy spreadsheets, reports and documentsTurn cross-system retrieval, reconciliation and drafting into a repeatable digital employee.If software automation must be built behind it, shorten implementation and maintenance.
数据不能出域以权限、审计、VPC/私有化和人工确认进入高敏业务。保护代码资产，并与内部 IDE、仓库、流水线和规范融合。Data cannot leave the boundaryUse permissions, audit, VPC/private deployment and human confirmation.Protect source code and integrate with IDEs, repositories, pipelines and standards.
经验依赖资深员工将岗位方法封装成 Skill，让新人和跨部门团队直接复用。让 AI 理解历史工程，生成项目文档、Rules 与可维护代码。Know-how sits with senior staffPackage role know-how as reusable Skills for onboarding and cross-team use.Let AI understand legacy projects and generate documentation, Rules and maintainable code.
研发周期与缺陷压力大用于研发 PM、材料、信息同步和跨团队协同。重点证明工程理解、代码生成、Debug、测试和 Review 的量化收益。Release pressure and defect loadSupport R&D PM work, documentation, updates and cross-team coordination.Prove measurable gains in understanding, generation, debugging, testing and review.

来源与方法

Sources and method

01 · WorkBuddy Financial Edition

100+ financial institutions; workbenches, deployment modes and quantified workflows. Open source

02 · Shanghai Jahwa / WorkBuddy

Usage mix, nine departments and supply-chain results. Open source

03 · Shenzhen Port Group / WorkBuddy

Seven scenarios, private boundary and workflow outcomes. Open source

04 · AI employees enter the workplace

JinkoSolar, Mengniu, Shanghai Jahwa and 1Drugstore adoption signals. Open source

05 · CodeBuddy enterprise practice

50k+ Tencent developers, top-three scenarios and workflow guidance. Open source

06 · TCL × CodeBuddy

Legacy code, migration, debugging and enterprise deployment metrics. Open source

07 · CodeBuddy enterprise overview

Vanke Onewo, DreamSky, China Merchants Bank and overall adoption. Open source

08 · Official CodeBuddy product site

IDE, Plugin, CLI and zero-to-one development journeys. Open source

方法说明：本次先检索了当前授权的司内版 IMA；仅能看到用户自己的空知识库，没有共享或订阅的销售材料，因此没有引用未获授权的内部销量数据。随后基于腾讯官方产品页、腾讯云开发者社区客户材料及主流媒体报道，按“客户规模、案例重复度、用户行为、可量化 ROI、部署深度”五个维度进行归纳。案例指标属于特定客户，不应直接视为所有客户的平均效果。
Method: The currently authorized enterprise IMA was searched first; it exposed only the user’s empty personal knowledge base and no shared/subscribed sales materials, so no inaccessible internal sales data is claimed. The synthesis then used Tencent product pages, Tencent Cloud customer materials and mainstream reporting, evaluated across customer scale, case recurrence, user behavior, measurable ROI and deployment depth. Customer-specific metrics should not be treated as market averages.