China Construction Bank:
the AI strategy, sourced.

Last updated Nov 19, 2026 · 18 dated moves · 22 sources

CCB runs AI as a state-directed, in-house programme: its own financial large models (FLMs) built on open-source models since March 2023, a privately deployed DeepSeek-R1 model, its own cloud with a growing GPU share, and a scenario count that it reports every quarter, which rose from 79 in mid-2024 to more than 900 by September 2026 (the later counts are described as generative AI scenarios rather than FLM scenarios).

How is China Construction Bank using AI?

China Construction Bank (CCB), a Chinese state-owned bank with RMB45,631,818 million in total assets at the end of 2025, builds its AI in-house. Since March 2023 it has trained its own financial large models (FLMs) on open-source base models, and in early 2025 it privately deployed a financial model based on DeepSeek-R1 for internal use. The bank measures progress by the number of business scenarios its large models serve: 79 at June 2024, 193 at end-2024 and 398 at end-2025 (financial large model and large-model scenarios), then more than 600 generative AI scenarios at June 2026 and more than 900 at September 2026, as its reports and results releases state. Its most documented application is AI across the credit approval process: in 2025 it served more than 113,000 corporate customers, cut the average net approval period by over 30% and drafts more than 50,000 documents a month. Account managers use a personal AI assistant, which logged 34.63 million interactions from 30 thousand users in 2024; The Asian Banker reported in November 2026 that executive vice president Lei Ming said relationship managers who used the assistant (Bangde) more effectively recorded customer-asset growth 57 percentage points above the bank-wide average. In 2025 CCB spent RMB26,722 million on fintech, 3.61% of operating income, and, according to The Asian Banker, launched a three-year "AI Plus" plan in the first half of 2026. It keeps human review of generative AI output and says generative AI does not make real-time fraud-blocking decisions.

Lead bank

China Construction Bank Corporation

Total assets (group, as reported, Dec 31, 2025)

RMB45,631,818 million total assets (group)[1]

Charter

Chinese state-owned commercial bank listed in Shanghai and Hong Kong, supervised by the National Financial Regulatory Administration (NFRA) and the People's Bank of China; overseas subsidiaries include China Construction Bank (Asia) in Hong Kong, and CCB FinTech is the group's software development subsidiary

Headquarters

Beijing, China

Flagship platform

CCB financial large models (FLMs), ChatCCB, Agent platform and CCB Cloud

AI leadership

Zhang Jinliang · Zhang Yi · Lei Ming · Li Jianjiang · Han Jing · Sheng Liurong · Wang Lixin · Kai Yang · Jin Panshi

  • Scenario count is the headline metric: large models served 79 internal scenarios at June 2024, 193 at end-2024, 274 at June 2025, 347 at September 2025, 398 at end-2025, over 400 at March 2026, then more than 600 generative AI scenarios at June 2026 and more than 900 at September 2026; the basis changed from FLM to generative AI scenarios (CCB results releases and reports).
  • Own models: FLMs built since March 2023 from 100-billion-parameter-class open-source models, iterated 16 times by end-2024; a DeepSeek-R1-based financial model was privately deployed for internal use in early 2025 (Reuters, March 2025).
  • Credit approval: an "AI-assisted + expert decision-making" model served more than 113,000 corporate customers in 2025, cut the average net approval period by over 30%, and generates over 50,000 documents a month with a generation rate of up to 90% (2025 annual report).
  • Front line: the personal AI assistant for account managers recorded 34.63 million interactions from 30 thousand users in 2024; an AI problem-solving assistant for outlet staff covered 98.73% of outlets in 2025.
  • Infrastructure: CCB Cloud reached 568.36 PFlops (FP32) at end-2025, up 12.10%, with GPUs and other new computing power above 25.63%; fintech investment was RMB26,722 million (3.61% of operating income) and digital finance staff numbered 30,085 (7.95% of headcount).
  • Agents: an open-access Agent platform launched in the first quarter of 2026; by mid-2026 CCB had AI agents for customer risk identification and structured risk analysis, and its Digital Brain analytics system was consuming about six billion tokens a day (The Asian Banker, November 2026).
  • Governance: CCB told visiting bankers in November 2026, as The Asian Banker reported, that it keeps human review of generative AI outputs, puts customer-facing AI through additional approval, and that generative AI does not directly make real-time transaction-blocking decisions.
  • Policy frame: the bank formulated an "AI+" Work Advancement Plan to implement the State Council's "AI+" opinions (2025 annual report), and launched a three-year "AI Plus" plan in the first half of 2026 (The Asian Banker).
  • Leaders: Chairman Zhang Jinliang and Vice Chairman and President Zhang Yi; executive vice president Lei Ming oversees technology; the chief information officer post held by Jin Panshi ended in April 2025.

What is CCB financial large models (FLMs), ChatCCB, Agent platform and CCB Cloud?

CCB began building financial large models in March 2023. Its 2024 annual report says it selected 100-billion-parameter-class open-source large language models and pre-trained, fine-tuned and applied reinforcement learning to them with its own text data, giving a family of models (language, text-to-image, multimodal and code) on one technology base with a Model as a Service platform. Staff reach them through ChatCCB, an FLM chatbot; AI Toolbox, where employees build their own applets (more than seven thousand by the end of 2024); Code Interpreter for data analysis; and a vectorised knowledge base. In early 2025 the bank completed a private, internal-only deployment of a financial model based on DeepSeek-R1. In the first quarter of 2026 it launched an open-access Agent platform inside the bank. The compute sits on CCB Cloud, which reached 568.36 PFlops (FP32) at the end of 2025, with GPUs and similar new computing power above 25.63% of the total. CCB FinTech, a subsidiary, does the group's software development.[3][16][10][1][17]

How fast has China Construction Bank scaled its AI scenarios?

CCB reports the number of business scenarios its large models serve in every results release. The count rose from 79 at June 2024 to 193 at the end of 2024, 398 at the end of 2025 and more than 900 by September 2026. From mid-2026 the bank describes them as generative AI scenarios rather than FLM scenarios, so the later figures may be on a wider basis.[5][3][1][13]

As ofScenarios as reportedBasis in the sourceSource date
2024-06-3079FinLLM in-house business scenarios2024-09-26[5]
2024-12-31193FLM application scenarios2025-03-28[3]
2025-06-30274FLM in-house application scenarios2025-09-26[6]
2025-09-30347FLM business scenarios2025-10-30[7]
2025-12-31398Large-model scenario applications2026-03-27[2][1]
2026-03-31Over 400AI application scenarios2026-04-29[10][9]
2026-06-30More than 600Generative AI application scenarios2026-08-28[11][12]
2026-09-30More than 900Generative AI large-model application scenarios2026-10-30[13]

What has CCB done on AI, and when?

CCB's dated AI moves by year2023Oct 26Generative AI applied in custom…2024Sep 26Half-year 2024: FinLLM in 79 sc…Oct 23AI engineering deputy head sets…Dec 19CCB (Asia) joins the HKMA's fir…2025Mar 28DeepSeek-R1 financial model dep…Mar 282024 annual results: 193 FLM sc…Jul 29Alibaba Cloud cites AI projects…Sep 26Half-year 2025: 274 scenarios a…Oct 30Third quarter 2025: 347 busines…2026Mar 272025 results: 398 scenarios and…Apr 27Annual report details AI in cre…Apr 29First quarter 2026: open-access…Aug 28Interim 2026: more than 600 gen…Oct 14"AI Trader" adds a large model…Oct 30Third quarter 2026: more than 9…Nov 03The Asian Banker: Bangde users…Nov 12Enterprise-wide AI operations c…Nov 18Digital Brain at six billion to…
Figure 1 · CCB's AI moves by year, from the timeline below. Filled boxes are the current year. Every entry cites its source in the list at the end of the page.
  • Nov 18, 2026

    Digital Brain at six billion tokens a day[17]

    Hosting The Asian Banker's Beijing study tour, CCB presented its real-time data warehouse and Digital Brain, an LLM and agent system for natural-language analysis that had run for about six months and consumed about six billion tokens a day, and described human review of generative AI output.

  • Nov 12, 2026

    Enterprise-wide AI operations competition[15]

    CCB held the final of its 2026 enterprise-level intelligent operations contest, run since August under its "AI+" requirements, with 124 projects entered; executive vice president Lei Ming presented the awards.

  • Nov 3, 2026

    The Asian Banker: Bangde users grew client assets faster[18]

    The Asian Banker reported that executive vice president Lei Ming, who oversees technology, said relationship managers who used the Bangde AI assistant more effectively recorded growth in customer assets under their management 57 percentage points above the bank-wide average, and that AI-assisted coding contribution was up 50% on 2025.

  • Oct 30, 2026

    Third quarter 2026: more than 900 scenarios[13]

    CCB said generative AI large models covered more than 900 application scenarios across the group and that it used AI for customer risk identification and structured risk analysis.

  • Oct 14, 2026

    "AI Trader" adds a large model for repo trading[14]

    CCB upgraded its RMB bond repo "AI Trader", first launched in 2021, with a large model that reads bond names and counterparty collateral requests; the bank says it is the first institution in the interbank market to apply a large model to RMB money-market trading.

  • Aug 28, 2026

    Interim 2026: more than 600 generative AI scenarios and risk agents[11][12]

    CCB said it had more than 600 generative AI application scenarios and had developed AI agents for customer risk identification and business risk analysis on an intelligent risk management platform.

  • Apr 29, 2026

    First quarter 2026: open-access Agent platform[10][9]

    The first-quarter report says CCB launched an open-access Agent platform within the bank and applied AI in more than 400 scenarios across customer service, marketing, operations and risk control.

  • Apr 27, 2026

    Annual report details AI in credit approval[1]

    The 2025 annual report says AI across the credit approval process served more than 113,000 corporate customers in 2025, reduced the average net approval period by over 30% and generated more than 50,000 documents a month; it also records fintech investment of RMB26,722 million and 30,085 digital finance staff.

  • Mar 27, 2026

    2025 results: 398 scenarios and an "AI+" Work Advancement Plan[2][1]

    CCB reported that large-model technology served 398 scenarios across the group, that it had set up an enterprise-level AI technology system covering data, models, computing power and security under an "AI+" Work Advancement Plan, and that CCB Cloud grew 12.10% to 568.36 PFlops.

  • Oct 30, 2025

    Third quarter 2025: 347 business scenarios[7]

    CCB said its FLMs had cumulatively supported 347 business scenarios and that it was prioritising intelligent responses to frontline staff questions and "virtual experts" in wealth management.

  • Sep 26, 2025

    Half-year 2025: 274 scenarios and daily model use up 96.96%[6]

    The 2025 half-year report says FLM inference efficiency rose several-fold, daily average utilisation rose 96.96% year on year, 274 internal scenarios were served, intelligent coding was activated by 94.37% of head-office developers (74.83% monthly active), and CCB won The Asian Banker's 2025 Asia-Pacific Best AI/ML Model Management Platform Innovation Award.

  • Jul 29, 2025

    Alibaba Cloud cites AI projects with CCB[22]

    Reporting Alibaba Cloud's AI agreement with Standard Chartered, the South China Morning Post said it followed Alibaba Cloud AI projects with China Construction Bank and ICBC; the scope of the CCB work was not described.

  • Mar 28, 2025

    2024 annual results: 193 FLM scenarios and ChatCCB[3][4]

    The 2024 annual report says 168 FLM scenarios were launched in 2024 for a total of 193 covering about half of employees, the FLMs were iterated 16 times, more than seven thousand AI applets had been built, and the personal AI assistant logged 34.63 million interactions from 30 thousand account managers; CCB Cloud reached 507.72 PFlops.

  • Mar 28, 2025

    DeepSeek-R1 financial model deployed privately[16]

    At the 2024 results press conference, President Zhang Yi said CCB had completed the private, internal-only deployment of a large financial model based on DeepSeek-R1 earlier in 2025, Reuters reported.

  • Dec 19, 2024

    CCB (Asia) joins the HKMA's first GenAI sandbox cohort[20][21]

    China Construction Bank (Asia) was one of 10 banks selected for the Hong Kong Monetary Authority's generative AI sandbox, run on Cyberport's AI supercomputing centre with technical trials from early 2025.

  • Oct 23, 2024

    AI engineering deputy head sets out the LLM approach at Sibos[19]

    Kai Yang, deputy general manager of CCB's AI engineering department, said the bank was deploying large language models in customer scenarios, risk management and operations, starting with internal scenarios before customers.

  • Sep 26, 2024

    Half-year 2024: FinLLM in 79 scenarios, CCB Cloud above 500 PFLOPS[5]

    The 2024 half-year report says the FinLLM served 79 in-house business scenarios across six segments, retrieval-based generation cut the time to produce a customer financial analysis report for credit approval from several hours to minutes, "Your Personal AI Assistant" was rolled out to account managers, and CCB Cloud passed 500 PFLOPS with GPUs and other new computing power above 20%.

  • Oct 26, 2023

    Generative AI applied in customer service, research and account-manager scenarios[8]

    CCB's third-quarter 2023 release said it pushed forward the application of generative AI in scenarios such as intelligent customer service, investment research reports and an account manager assistant.

Where does CCB use AI today?

CCB's AI systems on the status ladder from announced to in productionANNOUNCEDPILOTROLLING OUTIN PRODUCTIONAI across the credit approval processPersonal AI assistant for account managers…ChatCCB, AI Toolbox, Code Interpreter and k…Open-access Agent platformIntelligent risk control system and risk ag…Anti-gambling and anti-fraud modelsRemote Intelligent Banking CentreAI problem-solving assistant for outlet sta…Intelligent coding on the agile R&D platform"AI Trader" for RMB bond repoAI in security operationsDigital Brain with governed data accessHKMA GenAI sandbox (CCB Asia)
Figure 2 · Where AI runs at CCB, and how far along. 12 of 13 named systems are in production; the rest are rolling out, piloted or announced. Status is as described in the cited source at its date.
SystemWhat it doesStatusUse case
AI across the credit approval process[1][3]Multimodal FLMs distil expert reviewers' knowledge into chains of thought, draft rating review opinions covering five modules within one minute and review opinions of more than ten pages with a generation rate of up to 90%; reviewers edit before submission. Served more than 113,000 corporate customers in 2025 and cut the average net approval period by over 30%.In productionCredit scoring & underwriting
Personal AI assistant for account managers (Bangde)[3][1][18]A ChatBot-based assistant on a large-model plus small-model architecture with AI scripts, profiling, customer search and the AI Xiaozhuge Agent; 34.63 million interactions from 30 thousand account managers in 2024, with vertical AI agents for customer insight and asset allocation added in 2025.In productionGenerative & agentic AI
ChatCCB, AI Toolbox, Code Interpreter and knowledge base[3][6]ChatCCB is an FLM chatbot for content generation, knowledge Q&A and search of trusted external sites; AI Toolbox lets staff build applets from prompts and internal systems; Code Interpreter writes and runs analysis programmes on uploaded files.In productionGenerative & agentic AI
Open-access Agent platform[10][17]Launched within the bank in the first quarter of 2026; CCB described continuing development of a unified agent platform on privately deployed AI infrastructure in November 2026.In productionGenerative & agentic AI
Intelligent risk control system and risk agents[1][12][6]FLMs, small models and rules combined in a risk control system with intelligent rating and Rule Intelligence; by mid-2026 AI agents for customer risk identification and structured risk analysis. An enterprise model risk management platform won an Asian Banker award in 2025.In productionModel risk management
Anti-gambling and anti-fraud models[1][17]New anti-gambling and anti-fraud models in 2025, including e-CNY models; AI helps develop detection rules that are then embedded in transaction blocking, while generative AI does not make real-time blocking decisions; graph computing flags suspicious relationships.In productionFraud detection
Remote Intelligent Banking Centre[1]Large models, video customer service and digital avatars in remote service; the centre handled more than 400 million customer inquiries across channels in 2025.In productionCustomer-facing chatbots
AI problem-solving assistant for outlet staff[1]A single portal answering frontline staff questions, covering 98.73% of outlets in 2025.In productionGenerative & agentic AI
Intelligent coding on the agile R&D platform[6][18]Code completion, code review and test-case writing across 13 R&D scenarios; 94.37% of head-office developers activated and 74.83% monthly active at June 2025. The Asian Banker reported AI-assisted coding contribution up 50% on 2025 by mid-2026.In productionAI-generated code & coding agents
"AI Trader" for RMB bond repo[14]A large model reads bond names, codes and market shorthand to extract counterparties' collateral needs, with small models and rules validating collateral and quoting; collateral range extended from rate bonds to credit bonds (announced October 2026).In productionTrading & capital markets
AI in security operations[1]AI applied to code security scanning, security alert analysis and phishing email drills (2025 annual report).In productionCybersecurity
Digital Brain with governed data access[17]LLMs and agent frameworks over governed enterprise data with a semantic layer, isolated execution and item-level permissions; about six billion tokens a day after about six months of operation; customer-facing AI needs additional approval.In productionAI governance (general)
HKMA GenAI sandbox (CCB Asia)[20][21]China Construction Bank (Asia) was in the HKMA's first generative AI sandbox cohort, with trials on Cyberport's AI supercomputing platform from early 2025.PilotThird-party & vendor AI

Use-case links open the tracker's rules-by-use-case matrix: which documents govern each system type.

What has CCB disclosed in numbers?

MetricValueAs of
Total assets (group)RMB45,631,818 million[1]Dec 31, 2025
Generative AI large-model application scenariosMore than 900[13]Sep 30, 2026
Large-model scenarios at end-2025398[1][2]Dec 31, 2025
FLM scenarios at end-2024193 (168 launched in 2024), covering about half of employees[3]Dec 31, 2024
CCB Cloud computing power568.36 PFlops (FP32), up 12.10%; GPUs and other new computing power above 25.63%[1]Dec 31, 2025
Fintech investmentRMB26,722 million, 3.61% of operating income[1]Dec 31, 2025
Digital finance personnel30,085 (7.95% of headcount)[1]Dec 31, 2025
Fintech patents granted4,847 (3,561 invention patents)[1]Dec 31, 2025
Corporate customers served by AI credit approvalMore than 113,000; average net approval period cut by over 30%[1]Dec 31, 2025
Credit documents generated by FLMsOver 50,000 a month on average[1]Dec 31, 2025
Personal AI assistant interactions34.63 million from 30 thousand account managers[3]Dec 31, 2024
Outlets covered by the AI problem-solving assistant98.73%[1]Dec 31, 2025
Remote Intelligent Banking Centre inquiriesOver 400 million[1]Dec 31, 2025
Digital Brain token useAbout six billion tokens a day[17]Nov 18, 2026
Head-office developers using intelligent coding94.37% activated, 74.83% monthly active[6]Jun 30, 2025
AI applets built by employeesMore than seven thousand[3]Dec 31, 2024

“In terms of digital finance, we advanced the “AI+” initiative, with FLMs empowering 398 application scenarios cumulatively.”[1]

Zhang Yi · Vice Chairman and President · Apr 27, 2026

“We established an enterprise-level AI technology framework, further applied AI to key areas such as wealth management, inclusive finance and risk management.”[1]

Zhang Yi · Vice Chairman and President · Apr 27, 2026

“recorded growth in the customer assets under their management that was 57 percentage points above the bank-wide average”[18]

Lei Ming · Executive Vice President · Nov 3, 2026

“LLMs that avoid bias are key. First, clean the data in the training process. It must be diverse. Second, go for internal scenarios, then broaden to customers. Last is monitoring all results and updating the models.”[19]

Kai Yang · Deputy General Manager, AI Engineering Department · Oct 23, 2024

“We will vigorously serve the nation’s goal of achieving greater self-reliance and strength in science and technology, strive to build a leading bank in technology finance, solidly advance our digital and intelligent transformation, and continuously improve the quality and efficiency of serving the real economy.”[1]

Zhang Jinliang · Chairman · Apr 27, 2026

Who runs AI at CCB?

Zhang Jinliang

Chairman of the Board (since March 2024)[1]

Zhang Yi

Vice Chairman, President and chief compliance officer (president since May 2024); announced the DeepSeek-R1 deployment at the 2024 results[1][16][18]

Lei Ming

Executive vice president (since August 2025), overseeing technology; cited the Bangde AI assistant's results, as reported by The Asian Banker in November 2026[1][18][15]

Li Jianjiang

Executive vice president and chief risk officer (since May 2024)[1]

Han Jing

Executive vice president (since May 2025)[1][18]

Sheng Liurong

Chief financial officer (since November 2022)[1]

Wang Lixin

General manager, Data Management Department (as presented in November 2026)[17]

Kai Yang

Deputy general manager, AI engineering department (as named in October 2024)[19]

Jin Panshi

Former chief information officer (March 2021 to April 2025; ceased to serve by reason of age)[1]

Which regulators govern CCB's AI?

The regulators around CCB's AICCBChina (NFRA / PBoC…3 documentsHKMA1 documentBasel Committee1 document
Figure 3 · The regulatory perimeter. Each authority links to its tracker page; the number is how many of its documents this page cites. Solid lines are the bank's direct supervisors; dashed lines are consumer, market and global standard-setters that reach it through its activities.
AuthorityWhy it matters hereDocuments
China (NFRA / PBoC / CAC)CCB's home regulators. The NFRA's 2024 data security measures require central management, pre-use review and monitoring of AI models, which bear on CCB's in-house FLMs and its private DeepSeek-R1 deployment; CAC rules apply to any public-facing generative AI service; the PBoC awarded CCB's mainframe core migration a first prize in its FinTech Development Awards in 2025. CCB's AI plan implements the State Council's "AI+" opinions.NFRA Data Security Measures (金规〔2024〕24号)
CAC Generative AI Interim Measures (Order No. 15)
PBoC FinTech Development Plan 2022-2025
HKMAChina Construction Bank (Asia) is an HKMA-authorised bank and was in the first cohort of the HKMA's generative AI sandbox announced in December 2024.HKMA GenA.I. Sandbox circular
Basel CommitteeCCB's enterprise data warehouse, risk data views and model risk management platform feed AI models used in credit and risk decisions, which engages the Basel Committee's BCBS 239 risk-data aggregation principles.BCBS 239

What does the public record suggest about CCB's AI strategy?

Analysis by BankingNewsAI from the sources cited on this page. Observations, not advice.

01

The scenario count changed its definition

CCB's headline AI figure moved from "FLM scenarios" (193 at end-2024, 398 at end-2025) to "generative AI scenarios" (more than 600 at June 2026, more than 900 at September 2026). The jump is steep and the basis is not explained in the releases, so the series is best read as direction rather than a like-for-like count.

02

Outcome figures exist for credit, less elsewhere

Credit approval is the one area with outcome data (113,000 customers, over 30% shorter approval period, 50,000 documents a month). Most other disclosures are adoption counts. The Asian Banker noted in November 2026 that CCB did not give a standardised measure of the financial benefit of its Digital Brain.

03

Open-source and domestic stack

CCB builds on open-source base models and its own cloud rather than a Western model vendor; the only model named in its public record is DeepSeek-R1, deployed privately. The rising GPU share of CCB Cloud (over 20% in mid-2024, above 25.63% at end-2025) is the capacity figure to watch.

04

No chief information officer named since April 2025

The 2025 annual report lists no successor to Jin Panshi as chief information officer; executive vice president Lei Ming fronts technology at results briefings. A named AI or technology chief would be the next governance signal.

Which leader rankings does CCB appear in?

8 of the computed leaders rankings include CCB, each ranked from the same sourced records (bank pages, filings, hiring snapshots, the agent tracker, the incidents register).

How does CCB compare with its peers?

Its peer group on this site, by business model and size:

How is China Construction Bank using AI?

CCB uses its own financial large models in credit approval (more than 113,000 corporate customers served in 2025), a personal AI assistant for account managers, ChatCCB and AI applets for staff, risk control and anti-fraud models, remote customer service, coding assistance and bond repo trading. It reported more than 900 generative AI application scenarios at the end of September 2026.

Does China Construction Bank use DeepSeek?

Yes. CCB said at its 2024 annual results in March 2025 that it had completed the private deployment of a financial large model based on DeepSeek-R1 earlier that year. Reuters reported that private deployment means the model is for internal use only.

What are CCB's financial large models (FLMs)?

They are models CCB has built since March 2023 by pre-training, fine-tuning and applying reinforcement learning to 100-billion-parameter-class open-source models with its own text data. They include language, text-to-image, multimodal and code models on one base, served through a Model as a Service platform.

How many AI scenarios does CCB have?

CCB reported 79 large-model scenarios at June 2024, 193 at the end of 2024, 398 at the end of 2025, more than 600 generative AI scenarios at June 2026 and more than 900 at September 2026. The later figures are described as generative AI scenarios, so the basis may differ from the earlier FLM counts.

How much does CCB spend on technology?

CCB's fintech investment was RMB26,722 million in 2025, 3.61% of operating income, according to its 2025 annual report. It had 30,085 digital finance staff (7.95% of headcount) and 4,847 granted fintech patents at the end of 2025.

Who leads AI at China Construction Bank?

Chairman Zhang Jinliang and President Zhang Yi set strategy; executive vice president Lei Ming oversees technology and cited the AI assistant's results in remarks reported by The Asian Banker in November 2026. Wang Lixin heads the Data Management Department. CCB FinTech, a subsidiary, carries out the group's software development. The former chief information officer, Jin Panshi, left the post in April 2025.

How does CCB govern generative AI?

CCB told visiting bankers in November 2026 that it keeps human review of generative AI outputs, subjects customer-facing AI applications to additional oversight and approval, and does not let generative AI directly make real-time transaction-blocking decisions. Its Digital Brain uses data-item-level access permissions and isolated execution environments.

Is CCB using AI agents?

Yes. CCB launched an open-access Agent platform inside the bank in the first quarter of 2026, built vertical AI agents for account managers' customer insight and asset allocation, and by mid-2026 had developed AI agents for customer risk identification and structured risk analysis.

  1. China Construction Bank Corporation Annual Report 2025
    China Construction Bank · Apr 27, 2026
  2. Stable Progress with New and Better Achievements: China Construction Bank reports operating results for 2025
    China Construction Bank · Mar 27, 2026
  3. Announcement of Annual Results 2024 (Annual Report 2024)
    China Construction Bank · Mar 28, 2025
  4. Improving Quality and Progressing Stably: China Construction Bank reports operating results for 2024
    China Construction Bank · Mar 28, 2025
  5. China Construction Bank Half-Year Report 2024
    China Construction Bank · Sep 26, 2024
  6. China Construction Bank Half-Year Report 2025
    China Construction Bank · Sep 26, 2025
  7. China Construction Bank Third Quarter Report 2025
    China Construction Bank · Oct 30, 2025
  8. China Construction Bank Reports Operating Results for the Third Quarter of 2023
    China Construction Bank · Oct 26, 2023
  9. Stable Progress with a Good Beginning: China Construction Bank released its operating results for the 1st quarter of 2026
    China Construction Bank · Apr 29, 2026
  10. China Construction Bank First Quarter Report 2026
    China Construction Bank · Apr 29, 2026
  11. Stable Progress with Higher Productivity: China Construction Bank reports 2026 interim results
    China Construction Bank · Aug 28, 2026
  12. 2026 Interim Results Announcement (presentation)
    China Construction Bank · Aug 28, 2026
  13. Stable Progress with Improved Quality and Efficiency: China Construction Bank Reports Third-Quarter 2026 Results
    China Construction Bank · Oct 30, 2026
  14. 银行间市场首家!建设银行"AI交易员"用上AI大模型 (CCB's AI Trader adopts a large model; in Chinese)
    China Construction Bank · Oct 14, 2026
  15. 建设银行举办2026年企业级智能运营竞赛 (CCB holds 2026 enterprise intelligent operations contest; in Chinese)
    China Construction Bank · Nov 13, 2026
  16. China Construction Bank says it launched internal financial model based on DeepSeek R1
    Reuters via Yahoo Finance · Mar 28, 2025
  17. How CCB and MEGVII are addressing the data and identity challenges of AI in banking
    The Asian Banker · Nov 18, 2026
  18. CCB's interim results point to sector recovery as NIM stabilises and growth returns
    The Asian Banker · Nov 3, 2026
  19. Sibos: GenAI will be more impactful than ecommerce and the internet, says panel
    Finextra · Oct 23, 2024
  20. HKMA picks 10 banks, 4 tech firms as Hong Kong evaluates commercial benefits of GenAI
    South China Morning Post · Dec 19, 2024
  21. HSBC, StanChart, Alibaba Cloud Among First Cohort of HKMA’s Gen AI Sandbox
    Fintech News Hong Kong · Dec 23, 2024
  22. Alibaba, Standard Chartered partner to accelerate AI adoption in banking sector
    South China Morning Post · Jul 29, 2025

When CCB moves on AI, you'll read it here first.

when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free

plus every regulation, bank and agent page update, the morning after · leave any morning