Artificial intelligence: concrete examples from everyday life to work

The most common examples of artificial intelligence you already have in your pocket: the filter that moves spam, the navigator that calculates when you will arrive, the face that unlocks the phone, the automatically generated subtitles, the suggested videos. At work the AI extracts data from the documents and summarizes the meetings; in the warehouse autonomous robots guide; in healthcare it proposes suggestions to the doctor; in the public administration sort certified mail. Each example has a precise limit, and it is often the manufacturer himself who declares it.

A definition is needed to decide what goes on the list. The AI Act calls the AI system an automated system that, for explicit or implicit objectives, deduces from the input it receives how to generate outputs such as forecasts, content, recommendations or decisions (Article 3 of EU Regulation 2024/1689). The examples below are divided by scope: for each you find what AI does, where it stops and the source that documents it.

How to Recognize a Real Example of Artificial Intelligence

Not all that is automatic is artificial intelligence. Recital 12 of the AI Act excludes from the definition systems based on rules defined solely by persons to perform operations automatically, and indicates as a distinguishing feature the inferential capacity: derive models or algorithms from the data and arrive at a result that no one has written step by step (EU Regulation 2024/1689). An Excel macro or a fixed-time thermostat remain out.

Istat draws a similar boundary in the survey of businesses. Among what does not count as AI lists industrial mechanical robots without artificial intelligence components, template-based word generators, automated email advertising, and traditional chatbots with pre-programmed responses (Istat, Enterprise and ICT 2025). If the assistant of a site answers only with sentences already written, it is an interactive menu. As you learn from the data explains the guide to Machine Learning; the complete picture is in Artificial intelligence: what it is and how it works.

Examples of artificial intelligence: the map by scope

The applications of artificial intelligence that you encounter most often, each with its main limit:

Examples of AI in daily life

Artificial intelligence in everyday life almost always works inside an app or device, without a label that signals it.

The Mail Spam Filter

Gmail alone moves suspicious emails to the Spam folder and, according to the official guide, the more messages signals like spam the more it recognizes similar ones (Gmail Help). The same guide explains what to do when the emails of people you know end up in spam: it is the limit of any classifier. In 2017, Google declared to block spam and phishing with an accuracy of more than 99.9% thanks to machine learning (Google Workspace).

The navigator who estimates the time of arrival

Maps combine real-time traffic with the historic traffic of each street to predict how long it will take. In 2020 DeepMind explained that neural network models developed with Google Maps improved the accuracy of real-time estimates by up to 50% in cities such as Berlin, Jakarta, S,o Paulo, Sydney, Tokyo and Washington (DeepMind). The same article recalls that rush hour changes from day to day and that accidents and closures complicate forecasting.

Unlocking the phone with your face

Face ID converts the face depth map into a mathematical representation, compares it to the recorded one and adapts to changes such as makeup or growing beard. Apple indicates a less than 1 in 1,000,000 chance of a random person unlocking the phone, but higher for twins, siblings and children under 13; face data remains encrypted on the device (Apple Support).

Suggested videos and posts

Social networks and video platforms decide the order of content with Recommendation Systems. The Digital Services Act obliges online platforms to explain in the general conditions the main parameters of these systems and the options to modify them; those of very large size must offer at least one option not based on profiling (EU Regulation 2022/2065, Articles 27 and 38). If the feed always proposes the same things to you, that option is the first place to look.

The Automatic Subtitles

YouTube creates subtitles with voice recognition. The official guide warns that they are generated by machine learning algorithms, that the quality varies from video to video and that errors in pronunciation, accents, dialects or background noises can misrepresent what is said; in the live automatic subtitles are available only in English (YouTube Guide).

The car that reads the signs and brakes itself

The EU regulation 2019/2144 makes it mandatory on new vehicles the intelligent adaptation of speed, which derives the limit by observing the signs and data of maps, the warning of driver fatigue and, on cars and light vans, the emergency braking that detects vehicles, pedestrians and cyclists (EU Regulation 2019/2144). The driver can exceed the suggested speed and shut down the system, and the fatigue data should be deleted immediately after processing.

Examples of AI at work: documents, meetings, emails

Among Italian companies with at least 10 employees using AI, the most widespread technology is the Extracting information from texts (Istat, Enterprise and ICT 2025): data from invoices and contracts, requests to be classified, long documents to be summarized, as the guide to analysis of business documents. Works well on recurring documents and mistakes with confidence when the text is ambiguous, so the extracted numbers should be checked: the reason is in the guide to hallucinations of AI.

Meetings are another example. The Google Meet 'Take Notes for Me' feature, also available in Italian, creates a document with summary, decisions and next steps and warns participants with an icon; the guide warns that the summary may be incomplete or inaccurate, that the recommended duration is at least 15 minutes and that the function handles only one language per meeting (Google Meet Guide). Other tools are in the guide to Transcript of Meetings.

For emails the AI prepares drafts to be reread (ChatGPT to write emails). Act 132/2025, effective October 10, 2025, adds two borders. Professionals can use AI only for instrumental and supportive activities, with prevalence of intellectual work, and must inform the customer about the systems used (Art. 13). The employer must inform employees when using automated systems that affect recruitment, duties, evaluation or surveillance (Art. 11). For a professional study the practical examples are in the page on AI for accountants.

In sales and customer support

The most visible examples are the assistants who respond on the site or on WhatsApp, the products suggested in e-commerce, the scores that sort the contacts to call back. Marketing and sales are also the business environment where Italian companies adopt AI most often, according to Istat.

The main limit is transparency. Since August 2, 2026, Article 50 of the AI Act requests that those interacting with an AI system be informed of this, except when it is evident from the context (EU Regulation 2024/1689). You can see an assistant at the opera in the demo of WhatsApp chatbots for businesses and an ordering of contacts in the one on the lead scoring. Examples of automation within a company, from emails to quotes, are collected in examples of AI automation in the company.

Factory and warehouse applications

In the definition used by Istat AI systems can be incorporated into devices: autonomous robots for automated warehouse management or assembly, drones for monitoring production or handling parcels (Istat). Alongside these is image recognition, which a line can use to locate the pieces to be discarded.

The limit here is the safety of people. The Machinery Regulation places safety components and machines with self-evolving behavior based on machine learning among the categories that require conformity assessment by third parties, when they ensure safety functions; the Regulation applies from 20 January 2027, and software that performs automatic functions without learning remains outside this rule (EU Regulation 2023/1230; Date of application: European Commission). Examples by sector on the page on AI for production and in the guide to AI for stock and inventory management.

In health care: suggestions to the doctor, decision to the doctor

AGENAS is building an AI platform for primary care that supports physicians with non-binding diagnostic-therapeutic suggestions and indications for prevention, with a focus on chronic patients (AGENAS). For 2026, the Agency planned a trial with 1,500 primary care physicians, lasting 12 months, on three use cases: basic diagnostic framework, chronicity management, prevention (AGENAS, December 2025).

The limit is written in the law. Article 7 of Law 132/2025 defines AI systems in healthcare as a support for prevention, diagnosis, treatment and therapeutic choice, leaves the decision always to the doctor, recognizes the patient's right to be informed and requests reliable, verified and updated systems periodically (Art. 7). For private studies, from appointments to reminders, there are the guide to the AI for Medical Studies and the page on Private Health.

AI in public administration: the example of PEC

The INPS uses a system that automatically classifies the thousands of PECs received every day and sorts them to the most suitable office to deal with them; the project has entered the 2021 World Top 10 of the IRCAI, the international research center on artificial intelligence of UNESCO (INPS). Reading, classifying, routing: it is the same pattern that is found in the mail of many companies.

Law 132/2025 sets the perimeter. In the PA the AI serves to make the activity more efficient and to reduce the time of the processes, ensuring knowability of the operation and traceability of the use; it has a instrumental and supporting function, and the person remains solely responsible for the measures (Art. 14).

The misunderstandings that come back in the AI examples

• "If it's automatic, it's artificial intelligence." Handwritten rules, macros and chatbots with fixed answers are not, either for the AI Act or for Istat. The difference changes obligations and expectations.
• "If the AI says so, it's right." Subtitles, summaries and extractions make a plausible mistake. Where the result has consequences, a person controls it.
• "Data stays in the app." Depends on the service: when you report an email as spam, for example, Google receives a copy of the message and can analyze it, as the Gmail Help. Before uploading customer data, read what the tool does with it (Business Data and AI).
• "AI decides in place of people." In health care, in the PA and in the professions Italian law says the opposite: support yes, decision no.

From an example to the first use in your business

The most useful examples to copy are those that resemble your everyday work. Before choosing, test them with two lists.

To move from examples to your own case, the AI Guide for Companies explains how to choose the first process, and the one on the Business Functions shows where you start in administration, sales and staff.

If among these examples you have recognized one that resembles a job of your company, Giallo Studio starts from a first free call to see if it holds on to your data and where to start.