Understanding AI use cases
Search for a familiar name, choose an application area or find your task through seven straightforward questions. Explore how the application works, which data it processes and how it is built. Then assess risks and identify suitable controls.
Search requires JavaScript. All use cases are listed below by application area.
- 47Use cases
- 11Application areas
We include only use cases with established names and evidence from practice, the market, standardisation or regulatory supervision. Our selection criteria
Start with your task and find a matching use case
Know what you want AI to do, but not what the application is called? Seven questions take you from your task to suitable use cases.
About 2–3 minutes. You can select multiple answers. Subsequent options fit your choices so far; “Not sure yet” leaves them open. Your answers stay on this page.
Enable JavaScript for the questionnaire or explore all use cases by field.
All use cases
Find a task from your working day and explore its architecture, data flows and the human role.
Use cases describe specific tasks. System design explains how the system performs the task; several technical archetypes may be combined.
Office work and knowledge management
7 use cases
Staff use AI to write, research, translate and document conversations. AI agents complete entire work steps independently.
-
Generative AI assistant for staffEnterprise generative AI assistant
Staff draft, revise, translate and shorten texts using a language model their organisation provides in a secured environment.
Generative assistanceProvides results for subsequent use
-
AI assistant in office applicationsEmbedded workplace AI assistant (copilot)
Staff invoke the assistant directly in email, word processing, spreadsheets and chat; it accesses mailboxes and file storage using their permissions.
Knowledge assistant grounded in sourcesAction permissions depend on integration
-
AI knowledge assistant (enterprise search)AI knowledge assistant, enterprise search (RAG)
The assistant answers questions from a defined document collection and shows which documents support the answer.
Knowledge assistant grounded in sourcesProvides results for subsequent use
-
Speech recognition and transcriptionAutomatic speech recognition and transcription
Speech recognition automatically transcribes dictation, conversations and other recordings.
Generative assistanceProvides results for subsequent use
-
AI meeting minutes and conversation documentationAI meeting notes, ambient scribe
From recorded meetings and consultation conversations, the system drafts minutes or a case note with an action list.
Generative assistanceAction permissions depend on integration
-
Public authorities and businesses automatically translate texts, websites and documents, often using their own glossary.
Generative assistanceAction permissions depend on integration
-
AI agentAI agent, agentic AI
An AI agent breaks a task into steps and completes it largely independently using tools, interfaces and its own permissions in business applications.
- EU AI Act, Art. 50 para. 1: Disclosure of AI interaction
Agent with tool accessCan act within applications
Documents and casework
4 use cases
AI captures incoming documents, redacts personal information, reviews large case collections and analyses many free-text responses together.
-
The system reads correspondence, invoices and forms, routes them to the responsible team and transfers their data into business systems.
Analysis and processingAction permissions depend on integration
-
Automated anonymisation and redactionAutomated anonymisation, PII redaction
The system finds names, addresses and other personal information in documents and suggests redactions for a person to approve.
Analysis and processingAction permissions depend on integration
-
E-discovery and technology-assisted review (TAR)E-discovery, technology-assisted review (TAR)
The system ranks large document collections by their likely relevance to an investigation or information request, making review easier.
Analysis and processingProvides results for subsequent use
-
Text analytics for feedback and consultation responsesText analytics of feedback and consultation responses
The system summarises large volumes of free-text survey, complaint or consultation responses, identifies themes and classifies each response.
Generative assistanceProvides results for subsequent use
Customer service, marketing and platforms
6 use cases
Businesses and platforms use AI to communicate with customers, adapt offers, prices and advertising, moderate content and generate images, videos and voices.
-
Customer service chatbots and voicebotsCustomer service chatbot, voicebot
A bot independently handles customers' or citizens' requests in chat or by phone and passes difficult cases to people.
- EU AI Act, Art. 50 para. 1: Disclosure of AI interaction
Analysis and processingAction permissions depend on integration
-
The system uses a person's behaviour and profile to select products, content or search results and determine their order.
Analysis and processingProvides results for subsequent use
-
Dynamic pricingDynamic pricing, algorithmic pricing
Online retailers and platforms automatically adjust prices to demand, costs, capacity and competitors' prices.
Analysis and processingAction permissions depend on integration
-
Personalised online advertising (behavioural targeting)Behavioural advertising, ad targeting
The system uses a person's behaviour and profile to select which advertisements they see.
Analysis and processingProvides results for subsequent use
-
Platforms automatically identify unlawful content or breaches of their rules and remove, restrict or refer posts for human review.
Analysis and processingCan act within applications
-
Image, video and voice generationGenerative media (image, video, audio)
The system generates or edits images, videos and audio from a short description, for example for advertising and communication.
- EU AI Act, Art. 5 para. 1 Upara. 1 point ba, Art. 5 para. 1a, 1b: Non-consensual intimate depictions
- EU AI Act, Art. 5 para. 1 Upara. 1 point bb: Child sexual abuse material
- EU AI Act, Art. 50 para. 2: Marking artificially generated content
- EU AI Act, Art. 50 para. 4: Labelling deepfakes and published texts
Media generationProvides results for subsequent use
Banking and insurance
5 use cases
Banks and insurers use AI to assess creditworthiness, detect fraud and money laundering, evaluate insurance risks and handle claims.
-
A model estimates repayment likelihood from payment and customer data, helping determine whether a person receives credit and on what terms.
- EU AI Act, Annex III no. 5 point b: Creditworthiness
Analysis and processingAction permissions depend on integration
-
A model assesses each payment for suspected fraud in real time and stops suspicious payments or refers them for review.
Analysis and processingCan act within applications
-
Money laundering detection and sanctions screeningAML transaction monitoring and sanctions screening
Banks search account and payment data for money laundering patterns and check all parties against sanctions lists.
Analysis and processingAction permissions depend on integration
-
Life and health insurance risk assessment and pricingAI-based underwriting and pricing in life and health insurance
Insurers estimate an applicant's mortality or illness risk to determine whether to accept them and what premium to charge.
- EU AI Act, Annex III no. 5 point c: Life and health insurance
Analysis and processingAction permissions depend on integration
-
AI-assisted insurance claims handlingAI-assisted claims handling
The system analyses claim documents and photographs, checks entitlement against the policy, flags inconsistencies and prepares settlement.
Analysis and processingAction permissions depend on integration
Employment and education
4 use cases
Employers and educational institutions use AI to screen applications, allocate and monitor work, supervise online exams and assess exam performance.
-
AI-assisted candidate screeningAI-based candidate screening
The system uses CVs and cover letters to identify the best match for a role and suggest whom the company should invite for interview.
- EU AI Act, Annex III no. 4 point a: Recruitment and selection
Analysis and processingAction permissions depend on integration
-
Software assigns tasks and shifts to workers, tracks what they do and evaluates their performance.
- EU AI Act, Annex III no. 4 point b: Employment decisions
Analysis and processingCan act within applications
-
Online proctoringOnline exam proctoring
The system monitors online exam candidates through camera, microphone and screen and flags possible cheating.
- EU AI Act, Annex III no. 3: Education and vocational training
Analysis and processingAction permissions depend on integration
-
Automated assessment of exam performance (essay scoring)Automated scoring of exams and essays
A model assesses written or spoken exam answers and awards marks, alone or alongside human examiners.
- EU AI Act, Annex III no. 3: Education and vocational training
Analysis and processingAction permissions depend on integration
Biometrics and emotion recognition
3 use cases
Faces, voices and other physical features can reveal identity. AI uses them to recognise people, verify identity or infer emotions.
-
Remote biometric identification (facial recognition)Remote biometric identification and facial recognition
The system compares faces in camera images or photographs with a database to identify unknown people.
- EU AI Act, Art. 5 para. 1 point e, f, g, h: Prohibited biometric practices
- EU AI Act, Annex III no. 1 point a: Remote biometric identification
Analysis and processingAction permissions depend on integration
-
Biometric verification and video identificationBiometric verification and remote identity proofing
The system checks whether a person is who they claim to be, for example by comparing a live image with their ID photograph.
Analysis and processingAction permissions depend on integration
-
The system infers a person's emotions or intentions from their face, voice or behaviour.
- EU AI Act, Art. 5 para. 1 point e, f, g, h: Prohibited biometric practices
- EU AI Act, Annex III no. 1 point c: Emotion recognition
Analysis and processingAction permissions depend on integration
Administration, justice and public safety
7 use cases
Authorities, police and courts use AI to select applications for review, value property, analyse investigations and mass litigation, identify languages, and recognise plates or events in video.
-
Risk-based review selection for applications and benefitsRisk scoring for benefit and grant applications
Authorities assess applications or payments for error or fraud risk and use this to select cases for closer review.
- EU AI Act, Annex III no. 5 point a: Public assistance benefits
Analysis and processingAction permissions depend on integration
-
Automated property valuation (AVM)Automated valuation model (AVM)
A model estimates residential property market values from property characteristics and comparable sales.
Analysis and processingAction permissions depend on integration
-
Police analysis platforms combine data from different sources and display links between people, incidents and cases as graphs, tables or maps.
- EU AI Act, Annex III no. 6: Law enforcement
Analysis and processingAction permissions depend on integration
-
AI assistance in high-volume court proceedingsAI assistance in courts for mass litigation
In high-volume proceedings, the system organises pleadings, extracts each case's features and prepares draft decisions.
- EU AI Act, Annex III no. 8 point a: Administration of justice
Analysis and processingAction permissions depend on integration
-
Language and dialect recognition in asylum proceduresLanguage and dialect identification in asylum procedures
From a short speech sample, the system estimates which language and dialect an asylum seeker is likely to speak.
- EU AI Act, Annex III no. 7: Migration, asylum and border control
Analysis and processingAction permissions depend on integration
-
Cameras capture vehicle number plates, and the system automatically checks them against registration or police search data.
Analysis and processingAction permissions depend on integration
-
Intelligent video analytics in public spacesIntelligent video analytics
The system detects specific events or objects in camera images and alerts a control centre without identifying people.
Analysis and processingAction permissions depend on integration
Healthcare and emergency services
3 use cases
In hospitals and dispatch centres, AI analyses medical images, warns of deterioration such as sepsis and supports emergency call handling.
-
AI-assisted medical image analysisAI in medical imaging
The system marks abnormalities on X-rays, CT, MRI or mammography images to help clinicians interpret findings.
- EU AI Act, Art. 6 para. 1, Annex I Section A: Safety component of an assessed product
Analysis and processingAction permissions depend on integration
-
Clinical decision support (such as early sepsis detection)Clinical decision support and early warning systems
The system continuously analyses vital signs, laboratory values and patient records and alerts the care team to a possible deterioration.
- EU AI Act, Art. 6 para. 1, Annex I Section A: Safety component of an assessed product
Analysis and processingAction permissions depend on integration
-
AI support for emergency call handlingAI-assisted emergency call triage
The system listens during an emergency call and alerts dispatchers to signs of an emergency, such as cardiac arrest.
- EU AI Act, Annex III no. 5 point d: Emergency calls and emergency care
Analysis and processingAction permissions depend on integration
Manufacturing, retail and logistics
3 use cases
In manufacturing and retail, AI predicts equipment failures and sales and inspects products for defects.
-
Predictive maintenancePredictive maintenance and condition monitoring
The system detects impending machine or equipment failure from sensor and operational data so maintenance can be planned in time.
Analysis and processingAction permissions depend on integration
-
Visual and acoustic quality inspectionAI-based visual inspection
Cameras or microphones capture components and products on the production line; a model checks defects, completeness and correct variants.
Analysis and processingAction permissions depend on integration
-
Sales and demand forecastingDemand forecasting and automated replenishment
Retailers use a model to predict how much of each item each store will sell, then turn those forecasts into suggested orders or completed orders.
Analysis and processingCan act within applications
Robotics and vehicles
2 use cases
AI enables robots to move independently in warehouses and vehicles to drive on public roads, or supports human drivers.
-
Autonomous mobile robots in intralogisticsAutonomous mobile robots (warehouse robotics)
Robots independently move goods around a warehouse or accompany staff picking orders; a controller allocates orders and plans routes.
Analysis and processingActs on physical processes
-
Automated driving and driver assistanceAutomated driving and advanced driver assistance systems (ADAS)
Cameras and other sensors capture traffic conditions, and the system drives under defined conditions or assists the driver.
- EU AI Act, Art. 6 para. 1, Art. 2 para. 2, Annex I Section B: Vehicles, aircraft, ships and railway assets
Analysis and processingActs on physical processes
Software development and IT security
3 use cases
In development, AI suggests code or writes and tests it itself. In IT security, it detects attacks in networks and mailboxes.
-
Coding assistantAI coding assistant
The assistant suggests code, tests and fixes in the development environment for developers to review and accept themselves.
Generative assistanceProvides results for subsequent use
-
Coding agentAI coding agent
The agent takes on complete programming tasks, works in the code repository and terminal, executes commands itself and submits changes for review.
Agent with tool accessCan act within applications
-
AI-based cyberattack detectionAI-based threat detection
The system learns normal patterns in networks, devices, accounts and email traffic, then flags deviations that may indicate attacks, phishing or compromised accounts.
Analysis and processingAction permissions depend on integration
How to read the use cases
A use case describes the typical situation
It summarises what known deployments share rather than describing a particular product. The questions about your system help you see whether it works the same way.
More threats do not mean greater risk
Only your assessment establishes likelihood and severity for your system. A longer threat list does not make one use case more dangerous than another.
Controls are suggestions
Your project determines which controls you need and how to demonstrate their effectiveness.
AI Act notes state conditions
Each note states the condition under which a provision applies. Use the quick checker to establish whether it applies to your system.
Unknown does not mean No
Until you can answer a system question, the corresponding threat stays on your checklist.
Your own records determine your system's assessment
Known deployments and sources demonstrate that the use case exists. We derive checklists and notes from them. Only your own records show how your system actually works.
Cross-cutting threats
Review these 9 threats for every AI system, whatever its purpose. They are therefore grouped at the end of each detail page.
- Supply Chain: InfrastructureThe system relies on data centres, cloud services and suppliers' software. Their outages or attacks affect you too.
- Supply Chain: ModelsCheck the model's provenance, integrity and updates, even if you host it yourself.
- Supply Chain: DatasetsTraining, reference and test data often comes from outside and can be altered in transit.
- Adversarial InputsDeliberately altered inputs can cause incorrect model outputs.
- Insecure Output HandlingPrograms using the output often do not validate it. Incorrect or manipulated output then propagates unchecked.
- Application VulnerabilitiesThe software surrounding the model can also contain vulnerabilities.
- OverreliancePlausible outputs tempt people to accept them without thorough review.
- Model Drift & DegradationChanges in data, environment or use can worsen results.
- Middleware Exploits (AI Framework Attacks)Third-party software components underlying the system can also contain vulnerabilities.
Since 2 February 2025, all AI system providers and deployers must ensure that their staff have a sufficient level of AI literacy (Art. 4). Support the AI literacy of your staff
Our selection criteria
A use case is included only if it meets all four conditions.
Practical relevance
Established in practice, the market, standardisation or regulatory supervision
Clear name
A commonly used name in German or English
Traceable evidence
At least two independent, publicly accessible sources documenting real uses or expressly regulating the use case
Distinct risk profile
Distinct risks that set it apart from other use cases
Where the application area categories come from
| Field of use | Based on |
|---|---|
| Office work and knowledge management | Federal Statistical Office and Eurostat: natural language generation, speech recognition; OECD: human interaction support |
| Documents and casework | Federal Statistical Office and Eurostat: text mining; Plattform Lernende Systeme: intelligent automation |
| Customer service, marketing and platforms | OECD: personalisation, human interaction support; Bitkom: customer service, marketing |
| Banking and insurance | EU AI Act, Annex III, point 5; European Banking Authority (EBA), BaFin, European Insurance and Occupational Pensions Authority (EIOPA) |
| Employment and education | EU AI Act, Annex III, points 3 and 4 |
| Biometrics and emotion recognition | EU AI Act, Article 5 and Annex III, point 1 |
| Administration, justice and public safety | EU AI Act, Annex III, points 5–8; Dutch and UK algorithm registers |
| Healthcare and emergency services | DIN and DKE AI standardisation roadmap: medicine; EU AI Act, Annex III, point 5(d) |
| Manufacturing, retail and logistics | Plattform Lernende Systeme: predictive analytics, quality control; OECD, Annex A |
| Robotics and vehicles | Federal Statistical Office and Eurostat: autonomous machine movement; AI standardisation roadmap: mobility, industrial automation |
| Software development and IT security | BSI and French cybersecurity agency ANSSI on coding assistants; OWASP; Eurostat: IT security |
Frequently asked questions
- What is a use case?
- A use case is a common application of AI with an established name, such as credit scoring or a customer service chatbot. Each case includes known deployments, threats to review and AI Act provisions that may apply.
- What if no use case fits exactly?
- Choose the closest match and answer the questions about your system. If it performs several tasks, review each matching use case.
- Does the number of threats indicate the risk?
- No. It indicates how much to review. Only your assessment establishes a threat's likelihood and severity for your system.
- Is my system high-risk under the AI Act?
- That depends on its purpose; there is no blanket answer. Each note states the condition under which a provision applies. Classify your system in the Quick check; find the obligations in the Obligations compass.
Updated: 30 September 2026. Review checklist rules version 0.3.2, threat catalogue version 2026.10.24.2.