Featured Issues

The gaps in automation and how agentic AI fills themUpgrade from Airflow 2 to 3 (without breaking your DAGs)Data teams at WeWork, Glassdoor, and Foursquare are already on Airflow 3. Are you?Join the Airflow experts at Astronomer on December 11 at 11am ET/4pm GMT to see what’s new and get tips on planning your upgrade:DAG versioning, human-in-the-loop, and more!Native support for MLOps + GenAI workflowsBest practices for upgrading your DAGs (without breaking them)Attend live for a chance to win a physical copy of the Practical Guide to Airflow 3!Save your spotSponsoredWelcome to DataPro Issue 160.This week, we’re bringing you a powerful edition packed with expert insights from across the AI landscape. In our latest Packt Talks session, Abhishek sits down with Pablo Moreno, AI Product Manager at Board, to break down one of the most urgent challenges facing data and automation teams today: why traditional workflows keep breaking, and how agentic AI is emerging as the scalable, adaptable solution enterprises actually need.In this issue, we dive into a structured, problem-solving walkthrough of the episode, covering real-world use cases from pharma and HR, the rise of Crew AI, new AI-powered roles, and what the shift from deterministic automation to autonomous agents really means for practitioners. If you're navigating automation, building AI systems, or leading data strategy, this one’s loaded with takeaways you can apply instantly.💡Workshop Spotlight:Algorithmic Trading with Python (Cohort 2)Only 10 seats left. Join Jason Strimpel on December 12 for a hands-on workshop where you’ll design, test, and deploy algorithmic trading strategies in Python with pandas, VectorBT, and the Interactive Brokers API. Whether you’re leveling up as a quant or sharpening your finance toolkit, this live session covers the full workflow from finding edges to executing real strategies. Choose a Full Pass or VIP Pass, use FINAL20 for 20 percent off, and secure your spot before it’s gone.Book Now and Save 20% – Only 10 Seats LeftCheers,Merlyn ShelleyGrowth Lead, PacktSubscribe|Submit a tip|Advertise with UsWhen Automation Breaks: How Agentic AI Solves Hidden Problems in Packt Talks with Pablo MorenoEvery company believes their automation is working, until it suddenly breaks. A small formatting change, a new policy variation, or a shift in user behavior can turn even the most expensive automated workflow into a bottleneck. This is the silent problem most businesses face: systems built to save time often create more manual work in the long run.In the latest episode ofPackt Talks, our GLAbhishek Kaushikdives into this challenge withPablo Moreno, AI Product Manager atBoard, a leader in financial planning and performance management software. Together, they unpack why traditional automation struggles to keep up with modern complexity and howagentic AIis rewriting the blueprint for scalable, resilient operations. The conversation sets the foundation for a deeper exploration of real-world cases, emerging frameworks, and the future of human-AI collaboration, insights that shape the rest of this article.The Automation Myth: Why Your Workflows Still BreakTraditional automation has helped companies save time for decades, but its biggest weakness remains unchanged: it can’t adapt. When business conditions shift, when documents look different, or when people ask slightly different questions, automated systems often fail. This mismatch between rigid workflows and fast-changing environments has widened over the last decade. Agentic AI steps in as a new paradigm, one where automation doesn’t just follow rules but makes decisions within defined boundaries.TakeawaysAutomation saves time but fails in dynamic, unpredictable scenarios.The pace of change in the last 5–10 years has outpaced deterministic workflows.Agentic AI thrives where traditional automation breaks, handling complexity without endless reprogramming.77% of enterprises are exploring agentic AI, with 30% already seeing major reductions in manual work.Problem in Focus: When Rule-Based Automation Turns into Technical DebtMany organizations rely on hyperautomation, systems built on stacks of rules, templates, and RPA bots. But every new rule becomes a maintenance cost. A single formatting change can break an automated pipeline. In the pharma industry example Pablo Moreno shares, scanning marketing documents for legal disclaimers turned into an expensive cycle of bot retraining.TakeawaysRule-heavy automation becomes brittle and expensive at scale.Minor variations (orientation, typos, layout changes) often break RPA workflows.Over-reliance on deterministic rules leads to high maintenance debt.The real cost of automation isn’t deployment — it’s constant upkeep.How Agentic AI Fixes the RPA ProblemAgentic AI doesn’t rely on rigid patterns. It understands context, recognizes variations, and adapts to change. In the pharma compliance case, agents could detect legal phrases regardless of spelling errors, text rotation, or synonyms. Instead of adding rules, the system learned patterns. This shift replaced hours of manual review and repetitive bot tuning.TakeawaysAgentic AI extracts meaning, not formatting, reducing the need for rules.Agents adapt to document variance with little intervention.Maintenance drops because the system generalizes from examples.Compliance workflows become faster, cheaper, and more reliable.Why Crew AI Works: Solving the Complexity ProblemChoosing the right framework matters. Pablo highlights Crew AI as his preferred approach because it enables multi-agent processes while retaining simplicity. Unlike low-code/no-code tools that struggle once workflows become nonlinear or multi-context, Crew AI supports scalable architectures, evolving use cases, and quick iteration cycles.TakeawaysCrew AI enables multi-agent workflows without heavy engineering overhead.Low-code/no-code tools are fast but tend to break once complexity grows.Crew AI supports new agents, tools, integrations, and iteration loops.Scalability and adaptability should guide platform decisions, not speed alone.The HR Chatbot Challenge: When Reality Is MessyA multinational pharma company attempted to build an HR chatbot for 100,000 employees across 40 countries. The complexity was enormous: policies varied by 20–30 percent, acronyms differed across teams, and employee questions were full of local nuance. The breakthrough came from building a robust knowledge base, including an acronym dictionary, and extensively training the system to recognize contextual differences.TakeawaysReal-world enterprise AI must handle messy, inconsistent data.Local nuances in HR policies can’t be hard-coded into rules.Knowledge bases and acronym dictionaries significantly improve AI accuracy.Good framing and good inputs are critical for agentic AI performance.Continue reading the full article on the Packt Medium handle.See you next time!*{box-sizing:border-box}body{margin:0;padding:0}a[x-apple-data-detectors]{color:inherit!important;text-decoration:inherit!important}#MessageViewBody a{color:inherit;text-decoration:none}p{line-height:inherit}.desktop_hide,.desktop_hide table{mso-hide:all;display:none;max-height:0;overflow:hidden}.image_block img+div{display:none}sub,sup{font-size:75%;line-height:0}#converted-body .list_block ol,#converted-body .list_block ul,.body [class~=x_list_block] ol,.body [class~=x_list_block] ul,u+.body .list_block ol,u+.body .list_block ul{padding-left:20px} @media (max-width: 100%;display:block}.mobile_hide{min-height:0;max-height:0;max-width: 100%;overflow:hidden;font-size:0}.desktop_hide,.desktop_hide table{display:table!important;max-height:none!important}}

Mobile development blogs, tutorials and resources inside!Latest Mobile Dev Insights: iOS, Android, Cross-PlatformAdvertise with Us|Sign Up to the NewsletterMobilePro #201: iOS 26.2, Compose Multiplatform update, Amper release, and more…Hi ,Welcome to another week of MobilePro; this is edition no. 201.As we slide into the holiday season, the one time of year when no developer wants surprise fires to put out, Apple seems to be keeping things mercifully calm with iOS 26.2. The update is shaping up to land sometime mid-December, and it’s very much a “polish, not panic” kind of release. Smoother animations, tighter performance, and small UI refinements like the new Lock Screen opacity slider make up the majority of the update. This release hopefully contains nothing that should break your app. However, it’s worth giving your interfaces and widgets a quick once-over just to make sure they still look intentional on the new system.The quieter changes are happening inside the system apps and services your app may rely on. Updates to Reminders notifications, offline lyrics in Apple Music, and expanded translation and speech features in certain regions might affect apps that integrate with those APIs, especially in markets where users can even swap out their default voice assistant. None of it is major, but it’s the type of OS update that can create tiny side effects if you’re not paying attention. A small round of testing now means you get to enjoy your December without surprise bug reports trickling in while you're trying to disconnect for a bit.That’s not all the news this week. Let’s dive in.🛡️ App Store Accountability Act: A new federal proposal may shift age-verification responsibility from developers to Apple and Google nationwide.🎨 Compose Multiplatform 1.9.3 Update: Adds improved previews, shadows, context menus, and the new Material 3 Expressive theme across platforms.🧩 Amper 0.9.0 Release: Introduces extensibility previews, faster dependency resolution, and multiple quality-of-life IDE upgrades.📐 Material 3 v1.4.0 for Jetpack Compose: Delivers stable MD3 components, dynamic color support, and a new adaptive navigation suite.🤖 Claude Opus 4.5 Launch: Anthropic’s latest model brings major boosts in reasoning, coding, safety, and long-running agentic workflows.Stick around for this week’sDeveloper Tip to learn to build high-performance iOS appsandtheDid You Know? section to know more about Digital Markets Act for Apple Maps and Apple ads.AI-Powered Prototyping MasterclassYou can save 50% OFF on your seats - apply coupon code CLOUD50 at checkout.If you’re planning your next AI project, this is your window to lock in a proven framework to prototype your idea without coding! The AI Prototyping Masterclass is a 5-hour, hands-on session led by Michael Hyzy, an AI strategy leader who helps global organizations move from AI ambition to deployed, working products.We will demonstrate with Bolt.new, Figma, n8n, ChatGPT and Supabase. The best part? The workflow transfers across tools of your preference. Join us on December 6, and build your first AI prototype in just one day.Book Your Seat Now📱 What's Happening in Mobile Development?If there’s any major news in the world of mobile app dev in the last week, MobilePro has you covered.iOSApp Store Accountability Act would make Apple responsible for age verification: A new federal proposal, the App Store Accountability Act, would make Apple and Google (not individual developers) responsible for verifying users’ ages before app downloads. States like Utah and Texas already require this. Congress is now considering applying the rule nationwide. Supporters say it’s a “commonsense” way to protect kids and simplify verification, though Apple has pushed back against the shift.AndroidWhat's new in Compose Multiplatform 1.9.3: Compose Multiplatform 1.9.3 introduces major upgrades, including new @Preview parameters, customizable shadows, a new context menu API, and an experimental Material 3 Expressive theme. The release also brings iOS frame-rate controls, improved web support now in Beta, enhanced accessibility, and a new API for embedding HTML content. It updates multiple dependencies and adds tooling improvements like AGP 9.0.0 support and unified web distribution.Amper Update, November 2025, Extensibility Preview: Amper 0.9.0 introduces its first extensibility preview, allowing developers to create custom tasks and local plugins using Kotlin + YAML, along with a redesigned documentation website. The update also brings major performance gains in dependency resolution, new incremental Java compilation, Maven-like layout support, and multiple IDE enhancements like auto-sync and improved quick-fixes. This release sets the stage for broader plugin publishing and richer customization in future versions.Jetpack Compose Material 3 1.4.0 is out: Jetpack Compose Material 3 has been updated to version 1.4.0, bringing stable support for modern Material Design 3 components, dynamic color, and updated theming aligned with Android 12’s visual style. The release also includes a new 1.5.0-alpha09 adaptive navigation suite for developers building more responsive UIs.Kotlin introduces checker for unused return values: Kotlin 2.3.0 RC introduces a new experimental checker for unused return values, helping developers catch silent logic bugs, while stabilizing features like nested type aliases and exhaustiveness checks. The update also brings Java 25 support, improved Swift interop, and enhancements across Kotlin/Wasm and Kotlin/JS, including direct JS export of suspend functions. The stable release is expected next month or early 2026.Google's New LiteRT Accelerator Supercharges AI Workloads on Snapdragon-powered Android Devices: Google has launched LiteRT’s new QNN accelerator, delivering massive on-device AI performance boosts on Snapdragon 8–powered Android devices, up to 100× faster than CPU and 10× faster than GPU execution. Built with Qualcomm, QNN enables full NPU delegation for most models, unlocking real-time experiences like instant scene understanding using optimized vision-language models. Support is currently limited to select Snapdragon 8/8+ devices, with tools available via GitHub.Artificial Intelligence (AI)Introducing Claude Opus 4.5: Anthropic has released Claude Opus 4.5, a major upgrade that delivers state-of-the-art coding, reasoning, and agentic capabilities while significantly improving speed, efficiency, and safety. The model outperforms previous versions across engineering, research, vision, and math tasks, and includes strong defenses against prompt-injection attacks. Alongside the model, Anthropic rolled out platform updates, such as effort controls, better tool use, and enhanced Claude Code, Chrome, Excel, and desktop integrations to support more powerful long-running workflows.Looking to build high-performance iOS apps that stand out from the 1.6 million+ crowd on the App Store? Check out this guide: Mastering iOS App Development—Strategies for Success. It walks you through performance-optimization tricks (like efficient memory use, networking, UI/UX refinements and startup-time improvements) that help your iOS apps stay fast, smooth and user-friendly.In case you have any tips to share with your fellow mobile developers, do reply to this mail and we’d be glad to feature you in a future edition of MobilePro.💭 What is the Mobile Community Talking About?What are mobile app developers discussing? Do you have any concerns, advice, or tutorials to share?MobileProbrings them to you all in one place.How a proper MVVM + Repository setup cut 6 months of rework in a Flutter app: A team refactored their messy Flutter codebase, where UI widgets, API calls, and business logic were tangled together, into a clean UI → ViewModel → Repository/Service → (optionally) Use-Case architecture, and saw feature delivery times drop from 6–8 weeks to 5–7 days.Flutter + Serverpod brings full-stack CRUD power back to Dart/Flutter: The latest tutorial from Tech With Sam shows how to build secure CRUD operations for a task-management app using Serverpod as the backend. Thanks to shared Dart models, type-safe database access, and auto-generated client APIs, you get a production-ready Flutter + backend setup, complete with authentication, data validation, and clean error handling, without leaving Dart.How SwiftData persists AttributedString under the hood: The article reveals that SwiftData now supports storing rich-text AttributedStrings. but instead of a plain text field, it serializes the attributed data into a storable format (often via a wrapper or transformer), preserving styling and metadata for Core Data/SwiftData-backed storage. This means you can persist and reload styled, attribute-rich text (e.g. markdown, formatting, links) reliably in your app’s database.📚️ Latest in Mobile Development from PacktMobilePro presents the latest titles from Packt that ought to be useful for mobile developers.A perfect book for beginners with minimal coding experience who want to enter the world of Swift programming and iOS app development.🎉 Have fun building your first iOS app and start your iOS programming career🧠 Learn to integrate Apple Intelligence and the sleek new Liquid Glass UI for modern app experiences🧱 Establish a solid foundation with UIKit, testing, and deployment best practicesiOS 26 Programming for BeginnersBuy now at $40.49! The European Commission is reviewing whether Apple Maps and Apple Ads should be labeled as “gatekeepers” under the Digital Markets Act, which would subject them to stricter EU regulations because they reportedly have over 45 million monthly users.Apple, however, argues that both services have limited market share compared with giants like Google Maps, Waze and major ad networks, and should therefore be exempt from special treatment.Sourced from 9 to 5 Mac.👋 And that’s a wrap. We hope you enjoyed this edition of MobilePro. If you have any suggestions and feedback, or would just like to say hi to us, please write to us. Just respond to this email!Cheers,Runcil Rebello,Editor-in-Chief, MobilePro*{box-sizing:border-box}body{margin:0;padding:0}a[x-apple-data-detectors]{color:inherit!important;text-decoration:inherit!important}#MessageViewBody a{color:inherit;text-decoration:none}p{line-height:inherit}.desktop_hide,.desktop_hide table{mso-hide:all;display:none;max-height:0;overflow:hidden}.image_block img+div{display:none}sub,sup{font-size:75%;line-height:0}#converted-body .list_block ol,#converted-body .list_block ul,.body [class~=x_list_block] ol,.body [class~=x_list_block] ul,u+.body .list_block ol,u+.body .list_block ul{padding-left:20px} @media (max-width: 100%;display:block}.mobile_hide{min-height:0;max-height:0;max-width: 100%;overflow:hidden;font-size:0}.desktop_hide,.desktop_hide table{display:table!important;max-height:none!important}}