iOS 26 AI: How Developers Are Using Apple's Local AI Models

Apple's Foundation Models and Early App Integration

During the WWDC 2025 event held earlier in the year, Apple unveiled its Foundation Models framework.

This framework empowers developers to integrate the company's on-device AI models directly into their applications.

Benefits of Apple's Local AI Models

Apple emphasized that developers utilizing this framework benefit from access to AI capabilities without incurring any inference costs.

Furthermore, these locally-run models incorporate functionalities like guided generation and tool calling as standard features.

Impact of iOS 26 Rollout

With the widespread release of iOS 26, application developers have begun updating their offerings to leverage Apple’s local AI technologies.

It's important to note that Apple’s models are relatively compact when contrasted with prominent models developed by companies such as OpenAI, Anthropic, Google, and Meta.

Consequently, the initial implementations primarily focus on enhancing user experience and refining existing app functionalities, rather than fundamentally altering core workflows.

Early Adopters: Apps Utilizing Apple's AI

The following represents a selection of the first applications to integrate with Apple’s new AI framework.

• Details regarding specific apps will be added as they become available.
• Expect incremental improvements to existing features through AI integration.
• The focus remains on enhancing usability and providing a smoother user experience.

These early integrations demonstrate the potential of on-device AI to improve mobile applications.

Further development and wider adoption are anticipated as developers become more familiar with the Foundation Models framework.

Lil Artist: An Interactive Learning App

Lil Artist is an application designed to provide children with engaging, interactive learning opportunities. It focuses on developing a range of skills, including creativity, mathematical understanding, and musical appreciation.

Developed by Arima Jain and Aman Jain, the app recently received a significant update – iOS 26 – which introduced an innovative AI-powered story creator.

AI Story Creation Feature

With the new feature, users can initiate the creation of unique stories by first choosing a character and then selecting a desired theme.

The application then leverages artificial intelligence to generate a narrative based on these selections.

According to the developers, the text generation process within the story creator is driven by a model that runs directly on the device, ensuring privacy and efficiency.

This means the AI processing happens locally, rather than relying on a remote server.

Key Features and Benefits

• Interactive Learning: The app provides a dynamic environment for children to learn.
• Skill Development: Focuses on creativity, math, and music skills.
• AI-Powered Storytelling: Generates unique stories based on user input.
• Local AI Processing: Utilizes on-device AI models for enhanced privacy.

The integration of local AI models in Lil Artist demonstrates a growing trend among developers seeking to harness the power of on-device machine learning.

This approach offers benefits such as reduced latency and improved data security.

Daylish

Currently, the team behind the Daylish application is developing a new feature. This prototype aims to intelligently recommend emojis for events recorded in the user’s timeline.

Emoji Suggestions Based on Event Titles

The core functionality revolves around analyzing the title of each event entered into the daily planner. Based on this analysis, the system will automatically propose relevant emojis.

This feature is intended to enhance user experience within the Daylish app. It aims to make timeline entries more visually engaging and expressive.

Prototype Development

At this stage, the work is focused on building a functional prototype. This allows for testing and refinement of the emoji suggestion algorithm.

The developers are prioritizing accuracy and relevance in their emoji recommendations. The goal is to provide suggestions that genuinely reflect the event's nature.

Benefits of Automated Emoji Suggestions

• Improved visual communication within the app.
• Faster event logging, as users can quickly select suggested emojis.
• A more personalized and enjoyable planning experience.

The Daylish team believes this addition will contribute to a more intuitive and user-friendly experience for all its users.

MoneyCoach

The MoneyCoach application, a tool for managing personal finances, incorporates two innovative functionalities driven by on-device machine learning models.

These features enhance the user experience and provide valuable financial awareness.

Spending Insights

One key capability is the provision of personalized spending insights. For instance, the app can determine if grocery expenditures exceeded the typical amount for the current week.

This allows users to quickly identify potential areas for budget adjustment.

Automated Categorization

Furthermore, MoneyCoach leverages local AI to automatically propose relevant categories and subcategories when recording expenses.

This streamlines the data entry process, making it faster and more convenient for users to track their financial transactions.

The automated suggestions minimize manual effort and improve the accuracy of spending classifications.

LookUp Enhancements with Apple's AI

The vocabulary building application, LookUp, has recently integrated two innovative features powered by Apple’s artificial intelligence models.

A new learning function has been implemented, utilizing a locally-run model to generate illustrative examples for each word.

Furthermore, this mode prompts users to demonstrate their understanding by constructing a sentence showcasing the word’s appropriate application.

Word Origin Visualization

Beyond example generation, the application’s developer is also harnessing on-device AI capabilities to create a visual map detailing a word’s etymological origins.

This feature offers users a geographical representation of the historical pathways of language.

The Tasks Application and On-Device AI

Similar to several other applications, the Tasks app has integrated a functionality that automatically proposes tags for new entries. This is achieved through the utilization of locally running AI models.

These same models are also employed to identify tasks that occur repeatedly and subsequently schedule them for future execution.

Key Features and Capabilities

The application further empowers users by enabling voice input. The locally hosted AI model then processes spoken commands, dissecting them into individual, actionable tasks – all without requiring an internet connection.

This on-device processing ensures both privacy and responsiveness, as data doesn't need to be transmitted to remote servers.

Local AI models are proving to be a powerful tool for enhancing user experience within the Tasks app.

Day One Journaling App Integrates Apple's AI

The journaling application Day One, which is a property of Automattic, is now leveraging Apple’s on-device machine learning models. This integration provides users with automated highlights and intelligent title suggestions for their journal entries.

Furthermore, the development team has introduced a new functionality designed to stimulate more extensive writing. This feature generates prompts that encourage users to elaborate on their existing thoughts and experiences.

These prompts are dynamically created based on the content already present in the journal entry, aiming to facilitate deeper reflection and more detailed documentation.

Crouton

The Crouton recipe application is now leveraging the capabilities of Apple Intelligence. This integration allows for intelligent suggestions of tags relevant to each recipe.

Furthermore, Apple Intelligence is utilized within Crouton to automatically name timers, enhancing user convenience.

AI-Powered Recipe Breakdown

A key feature of Crouton’s AI implementation is its ability to dissect lengthy recipe text. It transforms this text into a series of concise, easily understood cooking steps.

This functionality aims to simplify the cooking process, making recipes more accessible to users of all skill levels.

Key Features Enabled by Apple Intelligence

• Recipe Tagging: AI suggests appropriate tags for improved recipe discoverability.
• Smart Timers: Timers are automatically named based on the cooking step.
• Step-by-Step Instructions: Complex recipe text is converted into clear, actionable steps.

By employing these features, Crouton seeks to provide a more streamlined and user-friendly cooking experience.

Signeasy

The digital signature application, Signeasy, is now leveraging Apple’s on-device machine learning models.

This integration allows the platform to identify and distill crucial information contained within contracts.

Contract Summarization

Users of Signeasy are now provided with concise summaries of the documents they are in the process of signing.

These summaries are generated by analyzing the contract's content using Apple’s local processing capabilities.

Benefits of Local Models

• Enhanced privacy, as data processing occurs directly on the user’s device.
• Faster processing speeds due to the elimination of network latency.
• Improved reliability, functioning even without an internet connection.

The implementation of Apple’s models represents a significant step towards streamlining the contract review process for Signeasy customers.

By offering readily accessible summaries, the app aims to empower users with a clearer understanding of their agreements before committing to a signature.

This feature is designed to increase user confidence and reduce the potential for misunderstandings.

Dark Noise: AI-Powered Soundscapes

The Dark Noise application, a popular background sound generator, has integrated Apple’s on-device machine learning capabilities. This allows users to create customized soundscapes through simple text prompts.

Generating Soundscapes with Text

Users can now describe the desired ambiance using just a few words. The app then leverages local models to generate a corresponding soundscape.

This innovative feature provides a new level of personalization for ambient sound experiences.

Customization and Control

Once a soundscape is generated, users retain full control over its composition. Individual elements within the soundscape can be adjusted to fine-tune the overall auditory experience.

This granular control ensures the soundscape perfectly matches the user’s preferences and needs.

Benefits of Local Models

Utilizing Apple’s local models offers several advantages. Processing occurs directly on the device, enhancing user privacy and reducing latency.

Furthermore, it allows for functionality even without an active internet connection.

Dark Noise’s implementation showcases a practical application of on-device AI, delivering a unique and user-friendly experience.

Lights Out

A new application, Lights Out, has been released for following Formula 1 racing and Grand Prix events. It was created by Shihab Mehboob, who is also known for developing the Twitter client Avery and the Mastodon client Mammoth.

Key Features

The core functionality of Lights Out centers around utilizing artificial intelligence models that operate directly on the user’s device. These models are designed to provide concise summaries of race commentary as events unfold.

This on-device AI processing allows for rapid summarization without relying on a network connection. It delivers key insights during a race in a streamlined manner.

Developer Background

Shihab Mehboob’s previous work includes the creation of popular social media applications. His development of Avery, a Twitter client, and Mammoth, a Mastodon client, demonstrates his experience in building user-focused applications.

With Lights Out, Mehboob is applying his expertise to the realm of motorsports, offering fans a novel way to stay informed during Formula 1 events.

The app aims to provide a focused and efficient experience for F1 enthusiasts. It leverages the power of AI to distill complex race commentary into easily digestible updates.

Capture's Innovative Use of On-Device AI

The note-taking application, Capture, is leveraging the power of on-device artificial intelligence. This implementation provides users with intelligent category recommendations.

These suggestions are dynamically generated as users input text into their notes or task lists, enhancing organization and efficiency.

Lumy

The Lumy application, a sun and weather-tracking tool, has been enhanced with artificial intelligence capabilities.

This integration allows Lumy to provide users with intelligent, weather-related recommendations directly within the app interface.

AI-Powered Suggestions

Utilizing AI, Lumy now proactively suggests actions and information relevant to current weather conditions.

These suggestions aim to improve the user experience by offering timely and helpful insights.

The implementation of AI demonstrates Lumy’s commitment to leveraging advanced technologies.

This is done to deliver a more personalized and informative experience for its user base.

CardPointers

CardPointers functions as a mobile application designed to monitor credit card spending. It also provides tailored recommendations for maximizing rewards points accumulation based on the user’s existing credit cards.

The latest iteration of the app incorporates AI technology. This allows users to pose questions directly regarding their credit cards and associated benefits.

Key Features and Functionality

With the integration of artificial intelligence, CardPointers offers a more interactive experience. Users can now receive instant answers to queries about their cards.

This new capability extends to understanding available offers and optimizing point earning strategies. The app aims to simplify credit card management.

The app’s core purpose remains expense tracking and rewards optimization. However, the AI-powered question answering feature represents a significant enhancement.

By leveraging AI, CardPointers seeks to empower users with greater control over their finances. It provides a convenient way to navigate the complexities of credit card rewards programs.

The inclusion of AI models allows for a more personalized and efficient user experience. This is achieved through natural language processing and intelligent data analysis.

Guitar Wiz

The Guitar Wiz application leverages the capabilities of the Apple Foundation Model framework to enhance the guitar learning experience.

Specifically, the app provides users with detailed explanations of chords as they progress through lessons.

Key Features & AI Integration

Guitar Wiz distinguishes itself by incorporating insights typically reserved for experienced guitarists, delivered at appropriate learning intervals.

Furthermore, the integration of the AI model significantly aids the developer in offering support for more than 15 different languages.

This broad language support expands the app’s accessibility to a global audience.

How the Apple Foundation Model is Utilized

• Chord Explanations: The AI provides clear and concise explanations of guitar chords.
• Advanced Insights: Players benefit from knowledge usually gained through years of experience.
• Multilingual Support: The model facilitates support for over 15 languages.

In essence, Guitar Wiz utilizes artificial intelligence to personalize and broaden the scope of guitar education.

The app aims to make learning guitar more accessible and insightful for players of all levels.

SmartGym

The SmartGym application leverages locally-processed Artificial Intelligence to transform workout descriptions into detailed, step-by-step instructions.

This includes specifying repetition counts, rest intervals, and necessary equipment for each exercise.

Key Features

Beyond workout generation, SmartGym provides users with comprehensive workout summaries.

These summaries detail monthly progress, offer breakdowns of established routines, and analyze individual exercise performance metrics.

Workout Generation Process

The core functionality of SmartGym centers around its AI-driven conversion process.

Users input a description of their desired workout, and the app intelligently translates this into a structured plan.

Progress Tracking and Analysis

• Monthly Progress: Users can monitor their overall fitness improvements over time.
• Routine Breakdowns: Detailed analyses of workout routines are provided.
• Exercise Performance: Individual exercise data is tracked to identify strengths and areas for improvement.

This data-driven approach empowers users to optimize their training regimens and achieve their fitness goals.

The app’s local AI processing ensures data privacy and responsiveness.

Stoic and Apple's On-Device AI

The Stoic journaling application is now leveraging Apple’s on-device machine learning models to enhance the user experience.

This integration allows for the delivery of uniquely tailored prompts to users, dynamically adjusted according to their recorded emotional state.

Key Features Enabled by Apple's Models

• Personalized Prompts: AI-driven suggestions are provided based on individual mood entries.
• Post Summarization: Users can quickly condense lengthy journal entries.
• Enhanced Search: The models facilitate efficient searching through past journal logs.
• Content Organization: Entries can be intelligently categorized and arranged.

By utilizing these capabilities, Stoic aims to provide a more insightful and organized journaling experience for its users.

The implementation of Apple’s local AI models allows for these features to be processed directly on the device, potentially improving speed and privacy.

SwingVision

SwingVision is an application designed to assist athletes in racquet sports – including tennis and pickleball – with refining their technique. This improvement is achieved through analysis of video recordings of their gameplay.

The developers of SwingVision are currently leveraging the power of Foundational models. These models enable the delivery of feedback that is both detailed and directly applicable to the player’s needs.

How it Works

The application analyzes video footage captured during play. It then identifies key elements of the player’s swing and form.

Using Foundational models, SwingVision goes beyond simple identification. It provides specific, actionable insights to help players correct flaws and optimize their performance.

Benefits of Using SwingVision

• Personalized Feedback: Receive tailored advice based on your individual technique.
• Actionable Insights: Understand exactly what you need to change to improve.
• Data-Driven Improvement: Track your progress and see the results of your efforts.
• Support for Multiple Sports: Applicable to both tennis and the rapidly growing sport of pickleball.

By utilizing advanced AI, SwingVision offers a powerful tool for players of all levels seeking to elevate their game. The integration of Foundational models represents a significant step forward in sports technology.

The app’s core function remains focused on providing players with the means to analyze and enhance their performance through video-based feedback.

Zoho

Zoho, a productivity software company originating in India, is now leveraging locally hosted models.

These models are being implemented to enhance functionalities such as summarization, translation, and transcription.

The integration spans across several of Zoho’s applications, including Notebook for document management and Tables for spreadsheet operations.

Implementation Details

The company is focusing on deploying models directly, rather than relying solely on external APIs.

This approach allows for greater control over data privacy and potentially reduces latency in processing requests.

Specifically, the local models are designed to improve the user experience within Zoho’s core productivity tools.

Applications Benefitting from the Technology

• Notebook: Enhanced document summarization capabilities are now available.
• Tables: Improved translation features are integrated into spreadsheet workflows.
• General: Accurate transcription services are being offered across multiple Zoho applications.

By utilizing these locally powered AI features, Zoho aims to provide a more efficient and secure experience for its users.

The move reflects a growing trend among software companies to bring AI processing closer to the point of data creation and consumption.

TrainFitness

The TrainFitness application leverages locally-run models to propose substitute exercises when specific equipment is unavailable to the user.

Equipment-Aware Exercise Suggestions

A key feature of TrainFitness is its ability to adapt workouts based on the user’s access to fitness equipment. This is achieved through the integration of on-device machine learning.

When a workout calls for an exercise requiring equipment the user doesn't possess, the app intelligently suggests a viable alternative.

How On-Device Models Work

Instead of relying on cloud-based processing, TrainFitness utilizes models that are executed directly on the user’s device.

This approach offers several benefits, including enhanced privacy and faster response times.

• Privacy: User data remains on the device.
• Speed: Suggestions are generated instantly, without network latency.
• Offline Functionality: The app functions even without an internet connection.

Benefits for Users

This feature provides a more flexible and accessible workout experience. Users are no longer hindered by a lack of specialized equipment.

TrainFitness empowers individuals to maintain their fitness routines regardless of their environment or available resources.

The system ensures that workouts can be completed effectively, even with limited access to traditional gym equipment.

Stuff

The to-do application incorporates a listening function, leveraging Apple's artificial intelligence models to transcribe spoken words into discrete tasks.

This list is continually being refined as further applications utilizing Apple’s on-device models are identified.

Functionality Details

The app’s listen mode provides a hands-free method for task creation. It directly translates user speech into actionable items.

Apple’s AI models power this voice-to-task conversion, ensuring efficient and accurate processing.

Ongoing Updates

Further applications that integrate Apple’s local AI capabilities will be added to this compilation as they are discovered.

The list is intended to serve as a dynamic resource for users interested in exploring these features. Regular updates are anticipated.

Key Features

• Voice-activated task entry.
• Utilization of Apple’s on-device AI.
• Continuous expansion of the app list.

The focus remains on identifying and documenting apps that benefit from Apple’s localized machine learning technology.

This allows for enhanced privacy and responsiveness compared to cloud-based solutions.