Cases
AWS
Real estate technology
Enterprise deployment
⚙ Technical reference
Deep 12/12
Level A
Altisource
Generate AI-driven modernization of Java legacy applications
Original challenge
Altisource needs to re-engineer its business applications for more than 10 years to support new real estate and mortgage operations, but the legacy of Java has a large code volume and its context spread across Jira, Bitbucket and design documents, and the traditional re-engineering speed is difficult to meet the business window.
Outcomes / progress
The AWS case revealed that team modernization exceeded 350,000 lines of Java code, with a 25 per cent increase in development productivity, four new applications delivered in four months and a 54 per cent reduction in code holes; the three-month average of Scrum story points went from about 160 to 200.
First-party enterprise source
Pharmaceuticals / preclinical research
Embedded deployment
⚙ Technical reference
Deep 12/12
Level A
Bayer
PRINCE: preclinical study search and a multi-agent research assistant
Original challenge
Bayer's preclinical data was fragmented across systems: structured study metadata sat alongside decades of PDF safety-study reports, while historical migrations left some metadata missing or inaccurate. Keyword search could not reliably answer questions spanning studies and documents, so researchers spent substantial time finding and checking the approved reports.
Outcomes / progress
A paper coauthored by Bayer and Thoughtworks reports that PRINCE integrates more than 18,000 internal studies. Among 15 to 20 frequent users surveyed, 75% reported significantly less time spent searching for information. After the multi-agent system was introduced, average response times for complex queries improved by 30%. The system's ability to fully meet user needs scored 3.1/5 on average; no drug-development timeline or financial return was reported.
First-party enterprise source
Banking
Enterprise deployment
Deep 12/12
Level A
BBVA
From employee pilots to an enterprise-wide generative AI adoption system
Original challenge
With more than 125,000 employees across multiple countries, BBVA needed to turn generative AI from a limited license pilot into a governed, reusable enterprise capability. The bank also had to manage use-case discovery, legal and compliance review, employee skills, and output quality—not merely count activated accounts.
Outcomes / progress
BBVA reports that more than half of its workforce uses generative AI weekly. The bank has identified more than 8,000 active use cases, about 700 of strategic importance, and estimates roughly three hours saved per employee per week. Its employee assistant handles more than 34,000 monthly queries over a knowledge base of 2,500+ documents.
Google Cloud
Private bank/asset management
Embedded deployment
Deep 12/12
Level A
Berenberg
Investment assistant, morning newspaper production and industry-wide AI stratification strategy
Original challenge
The portfolio manager at Berenberg needs to read and synthesize the voucher reports, company documents and market news on a daily basis, and study manual time limits for vomiting; as a regulated bank, generic AI must also access the proprietary investment framework and maintain compliance reviews.
Outcomes / progress
The case of Google Claude states that the production of process content such as Morning Mail has increased by 85 to 90%, saving about one hour per day per sale; the same team is able to cover the wider market.
First-party enterprise source
Financial technology
Enterprise deployment
⚙ Technical reference
Deep 12/12
Level A
Block
Goose general-purpose agent and G2 persistent workflow applications
Original challenge
Block wanted to extend AI beyond code assistance into HR, customer operations, legal, and other functions. A general chatbot could not safely read proprietary real-time data or take action, while non-technical employees could not wait for an engineering team to build every automation.
Outcomes / progress
Block reports a 25% reduction in manual work hours across more than 75% of its employee base. AI handles 65% of Cash App support cases; more than 90% of code submissions are partially or fully AI-assisted; and median weekly code changes increased 30%.
First-party enterprise source
Automotive component manufacturing
Enterprise deployment
⚙ Technical reference
Deep 12/12
Level A
Bosch
Synthetic defect data for stator-weld visual inspection
Original challenge
Automated optical inspection needs images covering many defect types, but Bosch’s manufacturing quality is high, so a new line cannot naturally accumulate enough defective parts. Waiting for real defects delays model and line commissioning; deliberately damaging parts is expensive and still may not cover the full defect space.
Outcomes / progress
Bosch reports that approximately 15,000 synthetic images were produced from a double-digit number of real images per defect type. The plant expects the approach to shorten the project by six months and generate annual productivity gains in the six-figure euro range. These are projected benefits while the new line is still being trained and expanded.
First-party enterprise source
Bank
Enterprise deployment
Deep 12/12
Level A
DBS Bank
All-line AI industrialization and personalization of clients
Original challenge
DBS needs to move from a long one-time delivery to a full-line capacity that can be replicated on a scale, governable, quantifiable economic value, taking into account bank risk, customer trust and employee adoption.
Outcomes / progress
The DBS 2024 report revealed that more than 1,500 models cover more than 370 cases, generating more than NZ$ 750 million in economic value throughout the year, and sending over 1.2 billion individualized reminders to more than 13 million clients.
First-party enterprise source
Digital health / software platform
Internal enterprise AI deployment
⚙ Technical reference
Deep 12/12
Level A
Doctolib
Unifying AI products through a Data & AI platform and a golden-path agent template
Original challenge
Doctolib originally had fragmented data, machine-learning, and engineering platforms, while AI product teams built their own evaluations, model access, and agent frameworks. The first AI-native products took several quarters to reach production. Teams spent time assembling infrastructure instead of product logic, and reuse between teams remained low.
Outcomes / progress
The Doctolib platform lead reports hundreds of evaluation experiments per day. Early AI-native products required several quarters to reach production, while one team at an internal hackathon reached a production beta with an agentic product in about 3 weeks. This single beta does not establish an average delivery time, and no patient outcomes, total ROI, or independent audit are reported.
First-party enterprise source
Local services / delivery platform
Enterprise deployment
⚙ Technical reference
Deep 12/12
Level A
DoorDash
Offline simulation and evaluation flywheel for an LLM support chatbot
Original challenge
After DoorDash replaced deterministic support decision trees with an LLM conversation system, a prompt change could improve one scenario while degrading another. Direct production testing would expose customers to risk; manual replay was slow and narrow; and edge cases such as fraud, high-value refunds, and extreme delays were especially difficult to validate.
Outcomes / progress
DoorDash reports that one major change reduced hallucinations by 90% in simulation and that the improvement carried into production. Iteration fell from days to hours; the system runs more than 200 multi-turn conversations in under five minutes; and the suite contains more than 50 evaluations.
First-party enterprise source
Mobility and local-services platform
Enterprise deployment
⚙ Technical reference
Deep 12/12
Level A
Grab
From a technical support bot to the enterprise LLM-Kit agent platform
Original challenge
Grab’s technical infrastructure team handled thousands of repeated support tickets in a six-month period. Its early support bot could answer questions, but every new agent team still had to rebuild authentication, secrets, deployment, observability, tool integration, and evaluation—making it difficult to move prototypes reliably into production.
Outcomes / progress
Grab reports that more than 500 services run on its internal agent framework, more than 50 remote MCP servers are registered, and the shared LLM gateway handles billions of tokens per month. Day-one production wiring that previously took at least two weeks now takes about one hour.
AWS
Electronic manufacturing services
Embedded deployment
⚙ Technical reference
Deep 12/12
Level A
Jabil
Global Manufacturing Data Base and Smart Workshop Assistant
Original challenge
Jabil has more than 100 manufacturing bases in more than 25 countries, machine data are permanently isolated from sites and cross-domain analysis is difficult; there are problems with performance, expansion and deployment efficiency in both workshop applications, and on-site personnel are dependent on scattered files and worksheets for troubleshooting information.
Outcomes / progress
The AWS case reported a 67-83 per cent reduction in deployment time, a 74 per cent reduction in data-processing time and a 23 per cent reduction in the cost of ETL after the use of Glue Flex; the first version of the smart workshop assistant was completed within one week, followed by gradual access to additional data sources.
IBM
Aerospace/defence
Embedded deployment
Deep 12/12
Level A
Lockheed Martin
Harmonized data base and security AI Factory
Original challenge
Lockheed Martin ' s data is scattered over multiple data lakes and 46 data management, analysis and business intelligence systems, and engineers have difficulty in quickly finding high-quality data and in building and managing AI applications on a scale that is securely constrained.
Outcomes / progress
According to the IBM client case, the number of data and AI tools was reduced by 50 per cent, 46 systems were replaced by an integrated platform with 216 data catalogue definitions automated; the question-and-answer pilot accuracy rate was raised from 45.45 per cent to 56.8 per cent after tip optimization, reaching 66.0 per cent after query optimization.
First-party enterprise source
Management consulting
Enterprise deployment
⚙ Technical reference
Deep 12/12
Level A
McKinsey & Company
Lilli enterprise knowledge and agent platform
Original challenge
Nearly a century of McKinsey knowledge was distributed across more than 40 internal sources, making discovery and synthesis time-consuming. The first platform version also depended on one model provider; as usage grew, cost, speed, and accuracy could no longer be optimized together.
Outcomes / progress
McKinsey reports 72% active adoption, more than 500,000 prompts per month, and up to 30% time savings in knowledge search and synthesis. The team grew from four people to more than 150.
IBM
Manufacture of medical equipment
Embedded deployment
⚙ Technical reference
Deep 12/12
Level A
Medtronic
Intelligent extraction of engineering drawings and spare parts
Original challenge
The large number and complex structure of engineering drawings for medical instrument products, and the reliance on manual reading and checking for key raw materials and spare parts identification, both slow product development and supply chain analysis, can easily lead to omissions due to differences in format and profile.
Outcomes / progress
The IBM case revealed that the accuracy of critical material information extraction was up to 90 per cent, the accuracy of spare parts number extraction was over 85 per cent, significantly reducing manual verification and saving several hours of processing; and the overall cost savings were not disclosed.
First-party enterprise source
Internet / enterprise compliance
Internal enterprise AI deployment
⚙ Technical reference
Deep 12/12
Level A
Meta
Expert-knowledge agent and verified correction loop for a specific compliance domain
Original challenge
In a specific compliance domain, experts must synthesize organizational positions, historical decisions, and external material. Important reasoning lives in scattered documents or in people's heads. Basic RAG still requires the model to infer rules from raw chunks on each run, making assessments slow, inconsistent, and difficult to improve safely after a human correction.
Outcomes / progress
Meta Engineering reports that after 3 sprints over about 6 weeks, individual assessments moved from days to minutes. Recipe-driven stages cut tokens per turn by about 80%; the team reports zero regressions across improvement cycles. Sample size, detailed timing baselines, test counts, and independent audit are not published.
First-party enterprise source
Digital banking / financial services
Internal enterprise AI deployment
⚙ Technical reference
Deep 12/12
Level A
Nubank
Five production customer-support agents built through evaluation-driven iteration and A/B rollout
Original challenge
Nubank handles varied support requests for card delivery, debt, credit limits, card management, and product explanations. A knowledge-base-only bot cannot reliably use live business data, perform controlled actions, or follow conditional workflows. Offline answer scores alone cannot establish whether real customer satisfaction and self-service outcomes improve.
Outcomes / progress
The paper reports 5 production agents. Against previous variants, the card-delivery agent improved AI transactional NPS by 37 percentage points and self-service rate by 29 percentage points, while remaining 10 points below expert human agents on NPS. Debt management remained 23.6 points below expert humans; the product-explainer agent had a temporary 1.5-point self-service decline. No complete ROI or independent audit is published.
Microsoft
IT services
Enterprise deployment
Deep 12/12
Level A
Presidio
From low adoption rate to organizational level Copilot productivity transformation
Original challenge
Presidio initially deployed Copilot to 300 sales, marketing, engineering, IT, management and project personnel, with a rate of implementation of only about 60 per cent; the simple distribution tool did not create stable skills, scenes and measurable business returns.
Outcomes / progress
The Microsoft case reported an increase from approximately 60 per cent to more than 90 per cent, with users saving an average of about 1,200 hours per month; a 90 per cent reduction in RFP processing time, project managers saving an average of six hours per week and a reduction in a three-day mission to two hours.
Microsoft
Air engine manufacturing
Embedded deployment
Deep 12/12
Level A
Rolls-Royce
Digital threads maintained for engine design, turbo leaf mass testing and prediction
Original challenge
The engine design cycle is year-by-year, with about 2 million cooling holes per month dependent on manual inspection and production bottlenecks; engine in-service maintenance also requires early detection of malfunctions from tens of thousands of parameters and long-term fragmentation of design, manufacturing and service data.
Outcomes / progress
The Microsoft client case reported that the leaf mass examination had resulted in a 30 per cent increase in machine utilization, a reduction in the number of days to near real-time failure processing; engine systems tracked over 10,000 parameters, and about 400 unplanned maintenance events were detected and avoided each year, saving millions of maintenance costs.
Microsoft
Clinical research services
Embedded deployment
⚙ Technical reference
Deep 12/12
Level A
Syneos Health
Clinical trial site selection, forecasting and document workflow
Original challenge
Syneos Health runs more than 500 clinical trials simultaneously, and test sites require manual collection and comparison of a large number of sites, subjects and historical data, often for months; the team also handles forecasts, files and client landscape discussions.
Outcomes / progress
The Microsoft case states that the platform was deployed over a nine-month period and that the initial site list could be generated from 24 to 48 hours, with subsequent screening reduced from several months to several weeks; and the 2024 site activation cycle reduced by approximately 10 per cent.
First-party enterprise source
Telecommunications
Enterprise deployment
⚙ Technical reference
Deep 12/12
Level A
Telstra
Enterprise AI across the data foundation, employee knowledge, customer service, and network automation
Original challenge
Telstra wanted to improve frontline knowledge access, customer self-service, and network reliability, but roughly 80 historical data platforms fragmented data and interfaces. Without a shared foundation, role-based training, and responsible controls, individual pilots would not scale into hundreds of business use cases.
Outcomes / progress
Telstra reports that AskTelstra supports more than 8,000 frontline employees and reduces average call time by over one minute. Its customer assistant nearly tripled the number of queries resolved without escalation. SmartFix performed 2.5 million proactive actions in FY25 and prevented nearly one million support calls.
First-party enterprise source
Telecommunications
Enterprise deployment
Deep 12/12
Level A
Vodafone
Business-level AI combination of passenger services, web carriers, employee assistants and software development
Original challenge
Large transnational telecommunications operators face high-frequency customer service dialogues, complex network failures, on-site services and software delivery efficiency problems at the same time, and it is difficult to develop cross-market reuse, harmonization of metrics and continuous adoption if teams are tested separately.
Outcomes / progress
Vodafone H1 FY26 investor material disclosed: Tobi processed approximately 60 million dialogues per month, SuperTOBi end-to-end resolution rate 70 per cent and NPS 8 percentage points higher; network repair time averaged 43 per cent; over 50,000 employees had a monthly implementation rate of more than 90 per cent; and 2,800 engineers developed a 12 per cent increase in productivity over their life cycle.
First-party enterprise source
Retail
Enterprise deployment
Deep 12/12
Level A
Walmart
Commodity Enterprise Trend-to-Product and Business Assistant Wally
Original challenge
Large retail commodity development and stock operations cross-trend judgement, design, procurement, door shop, channels and inventory data, traditional commodity business cycles, and it is difficult for merchant teams to explain in a timely manner the differences in the performance of single goods in different markets and channels.
Outcomes / progress
Walmart 2025 investor conference material stated that the first series of clothing supported by Trend-to-Product was 18 weeks shorter than the typical process and achieved strong sales performance; Wally had been used for cross-channel sales diagnostics by the comptoirs.
First-party enterprise source
Automotive manufacturing
Enterprise deployment
⚙ Technical reference
Deep 11/12
Level A
BMW Group
Factory Genius production-equipment maintenance assistant
Original challenge
Every minute of equipment downtime affects vehicle production. Maintenance staff must search equipment manuals, quality data, fault reports, planning files, and daily shift logs. Multiple plants were also building similar tools independently, limiting cross-site reuse.
Outcomes / progress
Factory Genius progressed from a plant pilot to an initial global application on BMW’s internal platform and can surface troubleshooting guidance within seconds. BMW does not disclose mean time to repair, avoided downtime, or financial benefit.
First-party enterprise source
Mobility platform / software
Enterprise deployment
⚙ Technical reference
Deep 11/12
Level A
Uber
Genie: an on-call engineering support and internal knowledge copilot
Original challenge
Teams across Uber ask roughly 45,000 questions each month in hundreds of Slack support channels. Answers are fragmented across Engwiki, an internal Stack Overflow, engineering requirement documents, and prior conversations. Repeated questions and multi-round waiting consume time for both internal users and on-call engineers.
Outcomes / progress
Genie is in production within Uber’s internal support workflow and includes traceable feedback, cost, and hallucination-evaluation loops. The engineering article describes the demand baseline and architecture, but does not report resolution rate, labor savings, or ROI.
Palantir
Aerospace
Typical Forward Deployed model
Deep 10/12
Level A
Airbus
Skywise Aviation Data Ecology and A350 Production
Original challenge
The fragmentation of data on aircraft design, production, suppliers and airline operations makes it difficult to develop industry-level shared operating systems.
Outcomes / progress
According to Palantir official sources, Skywise daily users exceed 50,000; and historical official sources, A350 production accelerates by 33 per cent and identifies cost savings of more than $1.7 billion per year.
OpenAI
Medical
Embedded deployment
Deep 10/12
Level A
Boston Children's Hospital
Automation of hospital operations + Diagnosis support for rare diseases
Original challenge
At the same time, hospitals face a great deal of duplication of operations and a very difficult combination of information on rare diseases.
Outcomes / progress
Officially, the 50+ automation saves approximately 60,000 hours, re-equipment labour costs $7M+ and helps to identify 40+ of previously unresolved rare diseases.
OpenAI
Technology/Software
Embedded deployment
Deep 10/12
Level A
Cisco
Codex Enterprise Software Project Closed
Original challenge
Large multiple warehouses, the C/C++ code library and strict security governance make it difficult for AI to code directly into production.
Outcomes / progress
Officially, 95 per cent + the new AI functionality is written by Codex; defects repair uplifts 10-15 times; and savings of 1,500 + engineering hours per month.
OpenAI
HR technology / SaaS
Enterprise deployment
Deep 10/12
Level A
HiBob
Internal GPT experiments on client products
Original challenge
The company wants employees to experiment with AI quickly, while transforming a truly effective internal prototype into a client-oriented product capability.
Outcomes / progress
Officially, 90 per cent + active use by employees; construction of 2,500 + experimental GPS, about 200 of which have successfully entered the internal workflow.
OpenAI
Travel/insurance
Enterprise deployment
Deep 10/12
Level A
Holiday Extras
Full AI productivity + Travel Insurance Assistant
Original challenge
Travel companies need both to improve the efficiency of their internal knowledge work and to extend AI to client insurance and personalized travel experiences.
Outcomes / progress
Officially, 95 per cent of the staff interviewed spent more than two hours a week, 92 per cent saved more than two hours a week, code debugging time fell by 75 per cent, cumulatively saved 500 + hours a week, or about $500k a year.
OpenAI
Finance
Embedded deployment
Deep 10/12
Level A
Morgan Stanley
Wealth Management Knowledge Assistant + Evals System
Original challenge
Wealth advisers need to quickly obtain credible answers from large and highly compliant research and process files.
Outcomes / progress
Officially, 98 per cent plus the use of a team of consultants; answerable materials expanded from approximately 7,000 questions to 100,000 documents; document accessibility increased from 20 per cent to 80 per cent.
OpenAI
IT services
Enterprise deployment
Deep 10/12
Level A
NTT DATA Group
Codex Complex Accident Analysis
Original challenge
Critical system incident analysis requires multiple senior engineers to collide logs, codes and context over a number of days.
Outcomes / progress
Officially, a three-day accident analysis for five senior engineers was completed by Codex in 30 minutes; Codex has approximately 9,000 active users.
Palantir
Aerospace manufacturing
Typical Forward Deployed model
Standard 9/12
Level A
Archer Aviation
eVTOL manufacturing and authentication data base
Original challenge
The manufacture and certification of new aircraft requires the continuous alignment of engineering, manufacturing, supply chains and authentication evidence and the separation of traditional systems.
Outcomes / progress
The official cooperation announcement stated that the parties were accelerating manufacturing on the AI basis; clients publicly quoted the emphasis that Ontology helped to integrate large-scale certification projects more quickly.
Palantir
Telecommunications
Typical Forward Deployed model
Standard 9/12
Level B
AT&T
S.C.O.U.T. Network Operations Application System
Original challenge
Data and operating processes for large telecommunications networks are extremely fragmented and require engineering teams to build business applications on a continuous basis on the same platform.
Outcomes / progress
Officially, S.C.O.U.T. grew from a joint project to 100+ AT&T full-time engineers for maintenance, with some 660 applications already available on Foundry.
OpenAI
Finance
Enterprise deployment
Standard 9/12
Level A
BNY
Enterprise Agent Platform Eliza
Original challenge
Large financial institutions need to allow large numbers of employees to build Agent themselves under security governance, rather than being developed by central teams on a case-by-case basis.
Outcomes / progress
Officially, the platform supports 125+ online usage, 20,000 employees are willing to build Agent, and legal review time has been reduced by 75 per cent.
OpenAI
AI developer tools
Enterprise deployment
Standard 9/12
Level A
Braintrust
Client needs directly preview branch
Original challenge
The cycle of client feedback between experiential products limits the speed at which the product team tries to mistreat.
Outcomes / progress
Officially, 50 per cent of the team moves to Codex within one month; clients request to form demonstrationable branches in minutes.
OpenAI
Automotive marketplace platform
Enterprise deployment
Standard 9/12
Level A
Cars24
Voice/dialogue Agent + service process automation
Original challenge
The loss of clues and service lead times has a direct impact on revenue, owing to the manual, fragmented and highly reliant dialogue in the trading links of the used vehicle platforms.
Outcomes / progress
Officially, AI Agent handles 1 million plus minutes of dialogue per month, increasing the client-service resolution rate by 50 per cent, reducing the lead time for critical services by 80 per cent, and recovering 12 per cent of previously lost vendor leads.
Palantir
Asset management
Typical Forward Deployed model
Standard 9/12
Level A
CAZ Investments
Investment trail screening and partner services
Original challenge
Investment teams review a large number of private fund-raising opportunities each year, while expanding partner services rather than linearly increasing human capacity.
Outcomes / progress
The Palantir network reports that the same resources can handle more than 100-fold leads and that lead processing time has fallen by more than 90 per cent.
OpenAI
Internet/media/game
Enterprise deployment
Standard 9/12
Level A
CyberAgent
ChatGPT + Codex Organization
Original challenge
Internet companies want to improve the quality of non-technical knowledge work and engineering development at the same time, and to get AI naturally into team decision-making.
Outcomes / progress
Officially, ChatGPT Enterprise has an active monthly usage rate of 93%.
Palantir
Aerospace/manufacturing
Typical Forward Deployed model
Standard 9/12
Level A
GE Aerospace
Fleet management and supply chain performance
Original challenge
Engine production and fleet support spans complex supply chains, spare parts, quality and maintenance data, with high delivery pressure.
Outcomes / progress
Official Q1,2026 material reported a 26 per cent increase in commercial and military engine output in 2025; Palantir listed AIP as a system to support fleet and supply chain performance.
Palantir
Consumer goods/food
Typical Forward Deployed model
Standard 9/12
Level A
General Mills
Supply Chain Smart Execution Project ELF
Original challenge
Four thousand suppliers, 200+ factories, approximately 1.2 million orders, and operators make about 50 million supply chain decisions per year.
Outcomes / progress
Approximately 3,000 purchase orders/days were assessed, with approximately 400 recommendations/days, approximately $40 K/day, $14M/year savings.
Palantir
Medical
Typical Forward Deployed model
Standard 9/12
Level A
HCA Healthcare
Carers' shifts match their abilities.
Original challenge
Traditional scheduling difficulties take into account both district needs, skill sets, employee preferences and future manpower needs, and data trails are incomplete.
Outcomes / progress
According to official sources, it has been expanded to nine acute inpatient facilities and plans are under way to reach 180+ hospitals with approximately 90,000 nurses.
Palantir
Consumer goods/food
Typical Forward Deployed model
Standard 9/12
Level A
Heineken USA
Distribution and transport supply chainAgent
Original challenge
Distribution, transport and supply chain plans need to address complex constraints with long development cycles in old ways.
Outcomes / progress
AIPCon publicly stated that it took the team three months to build the capacity that it took the last three years to complete.
OpenAI
Financial services / e-commerce
Enterprise deployment
Standard 9/12
Level A
Klarna
AI customer-service assistant
Original challenge
Large-scale multilingual payments and shopping services need to reduce waiting times while maintaining resolution rates and client satisfaction.
Outcomes / progress
During the first month of the online cycle, 2.3 million dialogues were processed, representing approximately two thirds of the total number of customer-service chats; the equivalent of 700 full-time sittings was reduced by 25 per cent for repeated consultations, with an average resolution time of 11 minutes < 2 minutes.
Palantir
Food/manufacture
Typical Forward Deployed model
Standard 9/12
Level A
Land O'Frost
Production schedule optimization
Original challenge
Production scheduling involves time-consuming manual organization taking into account orders, lines of production, switching lines, raw materials and delivery constraints.
Outcomes / progress
Official Business Update reports that the schedule has been reduced from about 40 hours to 30 minutes.
Palantir
Industrial spare parts/services
Typical Forward Deployed model
Standard 9/12
Level A
Parts Town
Passenger service and on-site service operations
Original challenge
Client support and on-site services for the industrial spare parts business require a rapid integration of products, orders, customers and service history.
Outcomes / progress
The official Q2,2026 submission stated that EBITDA margin, where project value opportunities are expected to exceed 200 basis points.
Palantir
Utilities
Typical Forward Deployed model
Standard 9/12
Level A
PG&E
Wildfire risk and public safety power outages (PSS)
Original challenge
Extreme weather conditions require the integration of meteorological, grid, customer and asset data over a very short period of time, precise determination of which lines are out of power and continuous notification to customers.
Outcomes / progress
According to Palantir official sources, the platform supports PSPS and wildfire risk mitigation; the company ' s overview reports a 99 per cent decline in the area affected by wildfires in 2022 compared to 2018-2020.
OpenAI
Retail/Technology/Finance
Enterprise deployment
Standard 9/12
Level A
Rakuten
Codex event response, CI/CD and autonomous development
Original challenge
In complex product ecosystems, failure recovery and development should be accelerated, and code review and safety standards should not be lowered.
Outcomes / progress
Officially, MTTR declined by about 50 per cent; some projects were compressed from quarterly to several weeks, with potential development increasing by three to four times.
Palantir
Enterprise software
Typical Forward Deployed model
Standard 9/12
Level A
SAP
SAP Cloud MigrationAgent
Original challenge
The large ERP migration consists of object mapping, code/configuration conversion, validation and extensive manual review, with high time and cost.
Outcomes / progress
The official Q1,2026 material stated that >99 per cent validation accuracy was achieved within two weeks of the early project and that it reduced the time and cost of migration by more than 70 per cent.
OpenAI
Technology/Software
Enterprise deployment
Standard 9/12
Level A
Simplex
Quantification of AI original software delivery
Original challenge
The system integration project needs to know exactly which link AI is designing, developing, testing and actually saving time, rather than deploying on the basis of a sense.
Outcomes / progress
Officially, single screen development time was reduced by 70 per cent, design by 40 per cent and internal integration testing by 17 per cent.
Palantir
Insurance
Typical Forward Deployed model
Standard 9/12
Level A
SOMPO Japan
Profit-making and sales decisions for commercial insurance
Original challenge
Commercial insurance sales and insurance require the translation of decentralized business data into enforceable client/risk decisions.
Outcomes / progress
The public client of the Palantir network quoted a profit improvement of about $60 million over the past three years and an increase of about $100 million is expected over the next three years; the related workflow extends to 10,000+ salesmen.
Palantir
Utilities
Typical Forward Deployed model
Standard 9/12
Level A
Southern California Edison
Network assets, weather and customer notification
Original challenge
Tens of millions of asset, meteorological and inspection records need to be harmonized, and customer notifications in the event of wildfires and electricity grid incidents must be fast and accurate.
Outcomes / progress
Official FY2022 Business Update discloses: 20M assets, 150M meteorological prediction points, 5M survey/vegetation records; client notifications are reduced from hours to minutes, with a 60 per cent reduction in leak notifications.
OpenAI
Construction/engineering
Enterprise deployment
Standard 9/12
Level A
Taisei Corporation
HR-led full-time AI talent development
Original challenge
Traditional construction enterprises need to expand AI from a small number of technical staff to full-time skills, while at the same time truly moving into day-to-day work.
Outcomes / progress
Officially, 3,300 Custom GPTs were created, with a weekly activity rate of 90 per cent and savings of more than 5.5 hours per employee per week.
Palantir
Medical
Typical Forward Deployed model
Standard 9/12
Level A
Tampa General Hospital
Patient flow / queue / Sepsis management
Original challenge
Hospital data are scattered and patient flow, staffing and clinical operations decisions require a single view in real time.
Outcomes / progress
Follow-up public information indicates that the length of hospitalization of Sepsis patients has been reduced by about 15 per cent.
Palantir
Government/agriculture
Typical Forward Deployed model
Standard 9/12
Level A
USDA
Agricultural projects are harmonized between Ontology and online distribution
Original challenge
Hundreds of legacy systems are fragmented, and farmers ' applications and project issuances require cross-system reconciliation of facts and status.
Outcomes / progress
According to official Q2,2026 materials, the project was open for 62 minutes, breaking the USDA online registration record and distributing more than $4.4 billion to farmers in the first five days.
OpenAI
Air/travel
Enterprise deployment
Standard 9/12
Level A
Virgin Atlantic
Legacy Reconfigure and Move App Quality
Original challenge
The aviation App online window is at high risk and it takes a significant amount of time for legacy code re-engineering and test coverage.
Outcomes / progress
The legacy re-engineering was reduced from about 2 weeks to about 30 minutes; the new App was close to 100 per cent single-measured coverage, with 0 P1 defects at the time of issuance.
OpenAI
Retail / e-commerce
Enterprise deployment
Standard 9/12
Level A
Wayfair
Commodity catalogue quality + automation of customer service for suppliers
Original challenge
The properties and labelling errors of millions of commodities affect search and experience; while suppliers support large volumes of work.
Outcomes / progress
Officially, 2.5 million commodity labels were amended, 41,000 vendor support sheets were automated per month and 1,200 ChatGPT Enterprise seats were deployed.
Palantir
Catering/supply chain
Typical Forward Deployed model
Standard 9/12
Level A
Wendy's Quality Supply Chain Co-op
Catering supply chain inventory and unusual disposal
Original challenge
Inventory and supply anomalies may have continued for days or weeks in the past, and teams need rapid location factors across supply chain data.
Outcomes / progress
Palantir publicly stated that the problem, which could have lasted several days/weeks, could be dealt with in about five minutes.
Palantir
Membership/non-profit
Typical Forward Deployed model
Standard 8/12
Level B
AARP
Rapid production of prototype membership services
Original challenge
Large member organizations need to quickly validate whether AI can improve its membership services and internal operations, while traditional projects have an excessively long project cycle.
Outcomes / progress
Official clients quoted the first prototype as being online within 45 days.
Palantir
Air/travel
Typical Forward Deployed model
Standard 8/12
Level A
American Airlines
Route Network Planning Ecology
Original challenge
Route network planning requires a complex trade-off between aircraft, airports, needs, moments and operating constraints.
Outcomes / progress
Official Business Update reports that tens of millions of dollars have been saved in about one year.
Palantir
Motorsport
Typical Forward Deployed model
Standard 8/12
Level A
Andretti Racing
RaceOS real-time car performance application
Original challenge
Track decision-making relies on telemetry, tactics and historical energy data for real-time racing vehicles and requires the rapid transformation of multiple analyses into operational applications.
Outcomes / progress
The Palantir structure document lists RaceOS as a representative case of the extended client standard AIP/Foundry architecture; no unified ROI is disclosed.
Palantir
Construction/manufacture
Typical Forward Deployed model
Standard 8/12
Level A
Associated Materials
Multi-business case / OTIF upgrade
Original challenge
The fragmentation of manufacturing and compliance data and the lack of a unified base for decision-making in multiple business processes.
Outcomes / progress
Palantir publicly stated that it had been on line 10+ within nine months, and OTIF had been raised from 40 to 90 per cent.
Palantir
Transport infrastructure
Typical Forward Deployed model
Standard 8/12
Level A
Cintra Ferrovial
Road incident real-time response
Original challenge
Highway accidents and anomalies require rapid detection and support of cross-sensor, traffic and operational data.
Outcomes / progress
Official clients cited the emphasis on real-time deployment to help both improve efficiency and save lives; undisclosed unified ROI.
OpenAI
Finance
Enterprise deployment
Standard 8/12
Level A
Commonwealth Bank of Australia
Full AI Capability + Customer Service/Anti-FraudAgent Extension
Original challenge
Large banks need to achieve universal access to AI while maintaining consistent governance in the context of security, data connectivity and high-risk client scenarios.
Outcomes / progress
Approximately 50,000 staff members were officially deployed to ChatGPT Enterprise; the subsequent expansion of Agent operations is still under way.
Palantir
Industrial manufacture
Typical Forward Deployed model
Standard 8/12
Level A
Eaton
Supply chain shortfall identification and disposal
Original challenge
Examples of 100k+ Daily Sales Order, 300+ Plant, 32M+ Spare Parts, 72+ERP, and the complexity of default and priority judgements.
Outcomes / progress
Palantir publicly reported a 25 per cent increase in productivity.
Palantir
Medical
Typical Forward Deployed model
Standard 8/12
Level A
Hospital for Special Surgery
Automation of insurance rejection claims
Original challenge
Insurance rejection claims require manual reading of materials, reconciliation of claims and preparation of claims, and individual claims take a long time.
Outcomes / progress
Official Business Update stated that approximately 45 minutes of manual work had been reduced to approximately 5 minutes, which should be constructed within 5 weeks.
Palantir
Engineering/industry
Typical Forward Deployed model
Standard 8/12
Level A
Jacobs
Plant energy consumption and dynamic maintenance
Original challenge
Industrial on-site energy consumption, chemical use and maintenance decisions are disconnected from sensor data.
Outcomes / progress
The pilot site achieved an annual reduction of about 20 per cent in total plant energy consumption over six months.
Palantir
Real estate/construction
Typical Forward Deployed model
Standard 8/12
Level B
Lennar
Enterprise data and AI field applications
Original challenge
The residential development process cuts across land, construction, sales and customer data and requires a front-line team to see directly the operational value of AI.
Outcomes / progress
Palantir client quotes a description showing that the business team saw significant results in about 14 minutes; the URI was not disclosed in the public material.
OpenAI
Finance
Enterprise deployment
Standard 8/12
Level A
MUFG
Full-time AI Assistant / Retail Finance Innovation
Original challenge
Large financial groups need to strike a balance between governance, security and large-scale employee adoption.
Outcomes / progress
Some 35,000 staff were deployed.
Palantir
Battery/manufacture
Typical Forward Deployed model
Standard 8/12
Level A
Panasonic Energy North America
Electrode Control Tower
Original challenge
The multistep judgement of the electrodes manufacturing process is highly artificial with on-site data viewing.
Outcomes / progress
The first was delivered in one month; a multi-step process of approximately four hours was reduced to approximately 15 minutes and material waste was reduced.
Palantir
Healthcare / CRO
Typical Forward Deployed model
Standard 8/12
Level B
Parexel
Regulating the generation and standardization of declaration material
Original challenge
Clinical/regulatory declarations require multiple documentation, data and expert duplication, long-cycle and difficult to standardize.
Outcomes / progress
The Palantir network publicly stated that the filing cycle could be reduced from an average of 10-12 weeks to approximately 3-4 weeks, a reduction of more than 50 per cent.
Palantir
Mining
Typical Forward Deployed model
Standard 8/12
Level A
Rio Tinto
Harmonizing data and value chain optimization for mine operations
Original challenge
Mining operations involve underground operations, equipment, personnel, security, supply chain and trade data, and systems are fragmented.
Outcomes / progress
The official cooperation bulletin emphasizes the location of projects such as the digitization of the Boratas value chain, underground operational data links and employee safety; no unified ROI is disclosed.
Palantir
Motorsport / manufacturing
Typical Forward Deployed model
Standard 8/12
Level A
Scuderia Ferrari
F1 power unit performance and reliability decision-making
Original challenge
Races and factories need to fast-track competitions, test stations, parts and power unit data to support performance and reliability judgements.
Outcomes / progress
The official press release stated that tasks that had taken several minutes to calculate in the past could be completed in a few seconds.
Palantir
Food/manufacture
Typical Forward Deployed model
Standard 8/12
Level B
Tyson Foods
ACE ~ dbt data platform migration
Original challenge
The migration of the old data platform to dbt required large-scale understanding, rewriting, compilation and review of the data model, which was originally estimated at 10-15 persons for approximately three years.
Outcomes / progress
Palantir publicly stated that the original estimate of 10-15 persons x 3 years was reduced to about 3-4 months for one person.
Palantir
Government/Public Service
Typical Forward Deployed model
Standard 8/12
Level A
U.S. Department of State
Medical qualification of diplomatic personnel
Original challenge
The pre-service medical review of diplomats involved a large amount of material, rules and cross-team processing and a long cycle.
Outcomes / progress
Official Business Update stated that the first three months of the application were on line, reducing the candidate ' s clarance cycle from 60 to 12 days.
OpenAI
Medical
Enterprise deployment
Standard 7/12
Level A
AdventHealth
Clinical administrative burden and document workflow
Original challenge
Clinical staff in the large hospital system spend a significant amount of time on recording, administrative and support tasks to reduce patient care time.
Outcomes / progress
Officially, some administrative tasks were reduced by 80 per cent.
Palantir
Catering/services
Typical Forward Deployed model
Standard 7/12
Level B
Aramark
Product Matching and Catering Data Classification
Original challenge
Data on products, formulations, supplies and sales are dispersed, and product matching and classification rely on a large number of people.
Outcomes / progress
About 99 per cent of the meal data categories are automatically classified within nine months.
Palantir
Energy / carbon management
Typical Forward Deployed model
Standard 7/12
Level A
bp
Oil well failure management
Original challenge
Oil well failures involve real-time equipment, maintenance, production and historical failure data, and traditional screening responses are slow.
Outcomes / progress
Official clients quoted a “three-digit” return on Palantir's investment.
OpenAI
Sports/car racing
Explicit FDE / embedded deployment
Standard 7/12
Level B
Chip Ganassi Racing
Match Day Data and Strategy Tool
Original challenge
With 200+ sensors, close to 1 billion data points per hour in the race, engineers need to quickly read and connect information during the race.
Outcomes / progress
Public cases emphasize improved decision-making speed and access to data, without giving a single ROI.
Palantir
Railways/logistics
Typical Forward Deployed model
Standard 7/12
Level A
CPKC
Data on railway operations and scalable decision-making platforms
Original challenge
The trans-regional railway network, involving complex real-time coordination of trains, people, goods and assets, requires a unified and scalable operating platform.
Outcomes / progress
The official AIPCon client quoted the emphasis on the future expansion of the platform; the specific ROI was not made public.
Palantir
Industrial manufacture
Typical Forward Deployed model
Standard 7/12
Level B
Cummins
Manufacturing dataOntology and operational decision-making
Original challenge
Although there is a large amount of manufacturing data, it is difficult for the business team to quickly access and act on the actual clients.
Outcomes / progress
Open client quotes highlight the fact that Ontology allows businesses to access data faster and more efficiently; ROI is not publicly harmonized.
OpenAI
Manufacturing/printing
Enterprise deployment
Standard 7/12
Level A
Dai Nippon Printing (DNP)
Cross-cutting process automation and processing upgrade
Original challenge
Large multi-purpose enterprises need to prove that AI is not a personal assistant, but can develop measurable process modifications in multiple sectors.
Outcomes / progress
Officially, 90 per cent of the cases produce measurable results, 100 per cent of the weekly work, 87 per cent of the associated automation rate of time cuts and a 10-fold increase in partial processing.
OpenAI
Materials/manufacture
Enterprise deployment
Standard 7/12
Level A
ENEOS Materials
Full-time knowledge and in-depth research in manufacturing enterprises
Original challenge
Material-manufacturing enterprises need to use proprietary information safely and reduce data aggregation and survey time in a context of stress and rising costs.
Outcomes / progress
Officially, 80 per cent of the staff of the pilot considered that work had improved significantly; HR data aggregation and analysis time had fallen by 90 per cent; and some surveys had been reduced from months to minutes.
Palantir
Insurance/claim settlement
Typical Forward Deployed model
Standard 7/12
Level B
ESI
Scanner understanding and compensation judgement support
Original challenge
There is a large number of poor quality scanned documents in the insurance operations, and manual extraction and judgement are time-consuming.
Outcomes / progress
Official clients quoted the team as having constructed the AIP module within approximately 90 minutes, which is of poor reading quality and gives better judgement.
Palantir
Technology/manufacture
Typical Forward Deployed model
Standard 7/12
Level B
Fujitsu
Forecasts, inventories and warnings
Original challenge
Operational data and machine learning capabilities are dispersed and it is difficult to access day-to-day operations directly.
Outcomes / progress
It is publicly stated that about $9M of annualized costs will be reduced over three months.
OpenAI
Legal/professional services
Enterprise deployment
Standard 7/12
Level B
Gilbert + Tobin
Corporate AI governance and scalability
Original challenge
While the legal profession wants to expand the use of AI, it also requires strict governance, confidentiality and professional responsibility.
Outcomes / progress
In September 2026, the case was made public; unapproved values were not added to the first edition.
Palantir
Defence/Industry
Typical Forward Deployed model
Standard 7/12
Level B
L3Harris
Decision-making on manufacturing and supply chain forecasting
Original challenge
Managers have no shortage of “what happened” watchboards, and no system to predict risk ahead of time and to give implementable programmes.
Outcomes / progress
Official clients cited the emphasis on retroactivity towards predictability and agility in decision-making, and public materials did not give uniform quantification of ROI.
Palantir
Medical
Typical Forward Deployed model
Standard 7/12
Level A
Nebraska Medicine
Discharge Lounge / Bed swing
Original challenge
The untimely release of beds after patients have been discharged has affected hospital capacity.
Outcomes / progress
It is publicly stated that the utilization rate of Discharge Lounge has increased by more than 2,000 per cent.
OpenAI
Government/Public Service
Embedded deployment
Standard 7/12
Level B
Polimill
Public AI infrastructure in Japan
Original challenge
Local governments need to share expert experience while meeting public sector governance and sustainable operations.
Outcomes / progress
In August 2026, the case was made public; the focus was on transforming the tacit knowledge of experts into a cross-autonomous reusable asset.
OpenAI
Electronic/manufacturing
Enterprise deployment
Standard 7/12
Level A
Samsung Electronics
Research and development, manufacturing, marketing and software development
Original challenge
Universal AI and Agent security need to be extended to complex global organizations.
Outcomes / progress
OpenAI described this as one of its largest corporate deployments.
OpenAI
Automobile/manufacturing
Enterprise deployment
Standard 7/12
Level A
Scania
Global Industrial Employees A.I.
Original challenge
Engineering and front-line teams are well-informed, but AI cannot stay at headquarters or in technical teams alone.
Outcomes / progress
Official cases state that there was a strong bottom-up approach to engineering and first-line and that there were early gains in productivity, quality and business processes; no ROI was disclosed.
Palantir
Retail/medical
Typical Forward Deployed model
Standard 7/12
Level B
Walgreens
Pharmacy-to-end AI workflow
Original challenge
Small-scale pilot projects were planned, and the operation of pharmacies required rapid expansion across shops, processes and systems.
Outcomes / progress
It was stated publicly that 10 pilot projects had been planned for 6 months and that about 8 months had been extended to about 4,000 shops.
Palantir
Retail
Typical Forward Deployed model
Overview 6/12
Level B
Lowe's
AI operational applications from POC to production
Original challenge
Retail scenery requires fast-tracking AI pilots into real production flows.
Outcomes / progress
Publicly it is stated that less than four months have passed from POC to production.
OpenAI
Retail
Enterprise deployment
Overview 6/12
Level B
Lowe's
Staff and client AI experience
Original challenge
Large retailing requires faster delivery of goods, services and in-house knowledge to employees and clients.
Outcomes / progress
OpenAI 2025 Enterprise AI report represents one of the cases.
OpenAI
Life sciences
Enterprise deployment
Overview 6/12
Level B
Moderna
Research and corporate knowledge
Original challenge
The pharmaceutical company hopes to extend generative AI to research, analysis and the daily work of staff.
Outcomes / progress
The OpenAI 2025 Enterprise AI reported it as a representative case.
Palantir
Medical
Typical Forward Deployed model
Overview 6/12
Level A
Mount Sinai
Clinical/operational process efficiency
Original challenge
Critical clinical operating processes are labour-intensive and limit turnover.
Outcomes / progress
Publicly stated 100% FTE efficiency improvement and expected new $13M income.
OpenAI
Semiconductor/Technology
Enterprise deployment
Overview 6/12
Level B
NVIDIA
Expanding organizational knowledge and professional competencies
Original challenge
Super-sized technical organizations need to expand their expertise, research and collaborative capacity to a larger number of staff.
Outcomes / progress
The client case was made public in August 2026.
Palantir
Railways/manufacturing
Typical Forward Deployed model
Overview 6/12
Level A
Trinity Rail
Stock optimization
Original challenge
Inventory occupancy versus cross-operational inventory decision-making is complex.
Outcomes / progress
Three months of construction, publicly stated, resulted in savings of approximately $30M and improved operating profitability.
Palantir
Air/travel
Typical Forward Deployed model
Overview 6/12
Level B
United Airlines
Prevention of delay in the operation of technology
Original challenge
Flight technology operations require rapid identification and coordination across data sources, with high delay/cancellation costs.
Outcomes / progress
Public statements avoided nearly 300 delays and 20 cancellations, corresponding to millions of dollars in cost avoidance.
OpenAI
Air/travel
Enterprise deployment
Overview 6/12
Level A
Virgin Atlantic
Client travel research and product planning
Original challenge
Client travel information, such as browsing, purchasing, valuing, taking advantage of, and feedback, is dispersed across systems and reports.
Outcomes / progress
Publicly claimed to have reduced the number of weeks of research to hours.
OpenAI
Network security/software
Enterprise deployment
Overview 5/12
Level B
1Password
Engineering development efficiency
Original challenge
Development, modification and validation of engineering time in a large code library.
Outcomes / progress
The OpenAI client case page reported a 21 per cent increase in engineering productivity.
OpenAI
SaaS/Services
Enterprise deployment
Overview 5/12
Level B
Intercom
AI customer-service assistant
Original challenge
Customer service needs to be balanced between quality, response speed and scale.
Outcomes / progress
OpenAI 2025 Enterprise AI report represents one of the cases.
OpenAI
Recruitment/platform
Enterprise deployment
Overview 4/12
Level B
Indeed
Recruit matching and job-seeking experience
Original challenge
The matching of large-scale jobs with job-seekers requires greater semantic understanding and individualization.
Outcomes / progress
OpenAI 2025 Enterprise AI report represents one of the cases.
OpenAI
Medical/insurance
Enterprise deployment
Overview 4/12
Level B
Oscar Health
Medical insurance operations and knowledge work
Original challenge
The insurance and medical operating process is a dense version and complex rule.
Outcomes / progress
OpenAI 2025 Enterprise AI report represents one of the cases.
Datawhale
Retail / e-commerce
Explicit FDE engagement
Deep 12/12
Level A
Anonymized provincial state power company subsidiary
Multi-line AI workflow and organizational capacity-building
Original challenge
Several lines of operation for electric power suppliers have continued to grow, but professional posts such as legal, financial and internal control have been limited by staffing constraints; a great deal of audit experience is in the minds of employees, and the base of AI has not actually entered the business stream.
Outcomes / progress
Datawhale reviewed the case and stated that the legal scene would save one legal production, covering the costs of approximately 10 person/years of labour in legal matters, finance, personnel, operations, technology, etc.
Datawhale
Finance
Explicit FDE engagement
Deep 12/12
Level A
The anonymous head fund.
AI Coding Research and Development Team Transformation
Original challenge
Developers have used AI coding tools individually, but tools, methodologies and competency standards are not harmonized, traditional PRDs are not adapted to AI collaboration, and individual effectiveness cannot be translated into organizational capacity.
Outcomes / progress
The Datawhale review of the case reported savings of about one third in code writing time and more than 50 per cent in document writing time.
Datawhale
Consumer goods/food
Explicit FDE engagement
⚙ Technical reference
Deep 11/12
Level A
Anonymity of consumer food products
Synergy in product development and packaging intelligent auditing
Original challenge
More than 100 SKUs are developed each year by enterprises and packaging clearance involves a wide range of design, regulatory, channel and product managers, and repeated checks create queue bottlenecks; R&D and channel experience is spread across the human brain and multiple systems.
Outcomes / progress
The formation of pre-packaging, product life-cycle knowledge sedimentation and business Builder culture programmes; no uniform quantitative benefits are disclosed on the Datawhale public page.
Datawhale
Manufacturing
Explicit FDE engagement
⚙ Technical reference
Deep 11/12
Level A
Anonymized auto parts and components supply chain enterprise
Transparency of manufacturing processes and unusual closures
Original challenge
While data on orders, purchases, inventories, shipments and refunds are all in ERPs, they are not linked to business logic; real business deviations are difficult to express, rely unusually on manual discovery, and the owners still have to ask around.
Outcomes / progress
The order was formed through full chain penetration, unusual proactive reminders and operating watchboards; no uniform quantification of benefits was disclosed on the Datawhale public page.
Datawhale
Government/Public Service
Explicit FDE engagement
Standard 9/12
Level A
Anonymous DMV.
Pre-screening of online operational materials
Original challenge
The online business materials are subject to manual checking, which is slow and repetitive.
Outcomes / progress
The public re-organization of the single review was reduced from about 15 minutes to 3-5 minutes.
Datawhale
Finance/business services
Explicit FDE engagement
Standard 8/12
Level A
An anonymous company of the United States.
Automation of financial processes
Original challenge
The processes of reimbursement, billing, audit, etc. are both semantic and subject to significant duplication of systems and high levels of accountability.
Outcomes / progress
The development of a transferable structure for “AI Sentencing + RPA Implementation + People Final Appeal”; some cases contribute to privatization numeracy inputs.
Datawhale
Manufacturing
Explicit FDE engagement
Standard 8/12
Level A
Anonymity Engineering Leasing Enterprise
Longtail demand acceptance / product matching
Original challenge
A large amount of long-tailing engineering requirements rely on manual judgemental equipment and programmes, with high cost of sales acceptance.
Outcomes / progress
The focus is on expanding the reach of long-tailed needs; public material is not available for harmonized indicators.
Datawhale
Manufacturing
Explicit FDE engagement
Standard 8/12
Level A
Anonymized German spare parts business
Teacher ' s experience transfer / newer training
Original challenge
Key process and troubleshooting experiences are found mainly in senior staff, and newcomers are highly dependent on teachers for their problems.
Outcomes / progress
The public material highlighted improvements in the development of newcomers and the sedimentation of experience; and the non-disclosure of harmonized financial indicators.
Datawhale
Retail
Explicit FDE engagement
Standard 8/12
Level A
Anonymized under-retail business
Door-to-door reconciliation.
Original challenge
Multiple shops, multiple payment channels and multi-system bills require manual checking by finance staff.
Outcomes / progress
The objective was to move from a paper-by-written check to an anomaly; public information was not disclosed as a unified ROI.
Datawhale
Manufacturing
Explicit FDE engagement
Standard 8/12
Level A
Anonymous industrial enterprises
Industrial procurement / material positioning
Original challenge
The SKU and material systems are complex, and procurement staff rely on the experience of older staff to locate replacement and correct material.
Outcomes / progress
Case focus shortens positioning time and reduces empirical reliance; public information does not disclose harmonized ROI.
Datawhale
Manufacturing
Explicit FDE engagement
Standard 8/12
Level A
Anonymous unmarked manufacturing enterprise
Pre-sale scheme and quotations
Original challenge
Pre-sale quotations relied heavily on engineers to understand non-target requirements and find similar items in historical programmes.
Outcomes / progress
Increased efficiency of outputs from the first draft of the programme; accurate values not disclosed in public information.
Datawhale
Consumer goods/retail
Explicit FDE engagement
Standard 7/12
Level A
An anonymous cross-border digestive enterprise
Sell-in / Sell-through business analysis
Original challenge
Headquarters shipments, channel sales and end-of-life sales have broken calibres, and fair statements do not answer real business questions.
Outcomes / progress
Establish an retrospective business analysis link, rather than simply a chat-type BI.
Datawhale
Construction/engineering
Explicit FDE engagement
Standard 7/12
Level A
Anonymized architectural enterprise
Diagram understanding and delivery aids
Original challenge
The drawings and design materials are complex and the process of extracting, checking and delivering information takes time.
Outcomes / progress
Reduction in time for duplicate inspections and documentation; indicators not made public.
Datawhale
Marketing/content
Explicit FDE engagement
Standard 7/12
Level A
Anonymized Local Life Enterprise
The short video content factory at the store.
Original challenge
A large number of stores need to produce low-cost short videos on a continuous basis, without scripting or production capacity.
Outcomes / progress
The focus is on reducing production costs and increasing production of individual content.
Datawhale
Life sciences
Explicit FDE engagement
Standard 7/12
Level A
Anonymous Biotech Enterprise
Digitalization of operating counters
Original challenge
Business and project data are scattered in Excel and documents, and it is difficult to query, aggregate and analyse operations.
Outcomes / progress
A common data base and question portal is formed; indicators are not publicly available.
Datawhale
E-commerce
Explicit FDE engagement
Standard 7/12
Level A
Anonymous cross-border electricity merchants
Inventory and replenishment
Original challenge
Multi-platform sales, warehousing and inventory data are dispersed and inventory judgements and replenishments rely on manual aggregation.
Outcomes / progress
Reduce the burden of data aggregation and manual judgement; specific indicators are not made public.
Datawhale
Logistics/supply chain
Explicit FDE engagement
Standard 7/12
Level A
Anonymous cross-border logistics enterprises
Optimization of container loading
Original challenge
The loading rate directly determines the profit, but the loading is highly dependent on manual experience.
Outcomes / progress
The case highlighted that loading thresholds such as 98 were directly related to profits; specific company indicators were not made public.
Datawhale
Trade/foreign trade
Explicit FDE engagement
Standard 7/12
Level A
Anonymous foreign trade enterprises
Automation of mail, quotations and product information
Original challenge
Foreign trade agents repeat inquiries, product information, quotations and correspondence.
Outcomes / progress
Reduce duplication of paperwork; specific indicators are not made public.
Datawhale
Telecommunications
Explicit FDE engagement
Standard 7/12
Level A
Anonymous telecommunications enterprises
Network Data Analysis / Smart Questions
Original challenge
Web logs and indicators are large and problem positioning and analysis reports rely heavily on analysts.
Outcomes / progress
The goal is to obtain insight at the minute level; public materials do not disclose uniform effects.
Datawhale
E-commerce
Explicit FDE engagement
Standard 7/12
Level A
Anonymous TikTok Electric Group
The Darfurian Screening and Building Union.
Original challenge
Finding, sifting, generating offers and continuing follow-up takes a lot of operational time.
Outcomes / progress
Expansion of the number of persons covered by a single person; disclosure of information without disclosing its precise effect.
Datawhale
Legal/professional services
Explicit FDE engagement
Standard 7/12
Level A
Group of anonymous lawyers
Preliminary draft for litigation retrieval / evidence collating / instrument
Original challenge
Counsel devotes considerable time to the search, preparation and initial draft of case files and duplicates.
Outcomes / progress
Reduction of desk work time; disclosure of material does not have a uniform quantitative effect.
Datawhale
Government/Public Service
Explicit FDE engagement
Overview 6/12
Level A
Anonymity Fireguard Company
digitizing paper records / warranty reports
Original challenge
Inspections, security records are extensive in paper-based documents and non-structured materials.
Outcomes / progress
Reduction in manual entry and paperwork; indicators not made public.
Datawhale
E-commerce
Explicit FDE engagement
Overview 6/12
Level A
Anonymized shoe servicer company
Commodity pictures and essay production
Original challenge
SKU is numerous and new, and the speed of production of photographs and commercial scripts has become a bottleneck.
Outcomes / progress
Enhancing content mass production capacity; specific ROI is not publicly available.
Datawhale
Government/Public Service
Explicit FDE engagement
Overview 5/12
Level A
Anonymized Urban Planning Agency
Urban planning research
Original challenge
Planning data, policies, reporting are fragmented and manual aggregation and research cycles are long.
Outcomes / progress
Reduction of desk analysis cycle; specific indicators not made public.