<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Piyush Bhatt</title>
    <description>The latest articles on DEV Community by Piyush Bhatt (@piyushpbhatt).</description>
    <link>https://dev.to/piyushpbhatt</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4215686%2F428f7872-4fcb-43eb-b365-5a42b6308a5f.png</url>
      <title>DEV Community: Piyush Bhatt</title>
      <link>https://dev.to/piyushpbhatt</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/piyushpbhatt"/>
    <language>en</language>
    <item>
      <title>AI Document Processing vs OCR: Which Do You Need?</title>
      <dc:creator>Piyush Bhatt</dc:creator>
      <pubDate>Thu, 12 Nov 2026 17:35:10 +0000</pubDate>
      <link>https://dev.to/piyushpbhatt/ai-document-processing-vs-ocr-which-do-you-need-21m5</link>
      <guid>https://dev.to/piyushpbhatt/ai-document-processing-vs-ocr-which-do-you-need-21m5</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://www.sharpquest.com/blog/posts/2026/october/ai-document-processing-vs-ocr/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-ai-doc-ocr" rel="noopener noreferrer"&gt;SharpQuest blog&lt;/a&gt;, written by Piyush Bhatt.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key takeaways&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;OCR recognizes text from an image. AI document processing understands the contextual meaning of the text and returns the specific field values your system requests, such as the company name, invoice number, total amount and due date.&lt;/li&gt;
&lt;li&gt;OCR with templates works well for one fixed form that never changes. When layouts vary by source (suppliers, customers, applicants), each layout needs its own template.&lt;/li&gt;
&lt;li&gt;AI handles layouts it hasn't seen, reads tables and line items, and gives a confidence level for each field.&lt;/li&gt;
&lt;li&gt;Neither is perfect: build in validation rules and a review queue so a person checks what the AI isn't sure about.&lt;/li&gt;
&lt;li&gt;Test on your own documents before deciding. That's the only accuracy number that matters.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If your team types data from PDFs, scans or email attachments into another system, you've probably been told to "use OCR". And OCR has been around for decades. So why is everyone now talking about AI document processing, and is it worth the change?&lt;/p&gt;

&lt;p&gt;The short answer: OCR reads the characters, AI reads the document and understands its meaning. Which one you need depends on how much your documents vary and what you need out of them.&lt;/p&gt;

&lt;h2&gt;
  
  
  What OCR does
&lt;/h2&gt;

&lt;p&gt;Optical character recognition (OCR) turns an image of text into text a computer can work with. Scan an invoice and OCR gives you the words and numbers on the page. Many older document systems are built on OCR, and they work best with documents in a fixed layout.&lt;/p&gt;

&lt;p&gt;What it doesn't give you is meaning. OCR doesn't know which number is the invoice total amount and which is the purchase order number. To get fields out, OCR tools rely on &lt;strong&gt;templates&lt;/strong&gt;: rules that say "the total is in this box, at this position, after the word &lt;em&gt;Total&lt;/em&gt;". Each new layout needs its own template.&lt;/p&gt;

&lt;p&gt;That works well when:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;You process &lt;strong&gt;one form you control&lt;/strong&gt;, such as your own application form or a government form with a fixed layout.&lt;/li&gt;
&lt;li&gt;You only need &lt;strong&gt;searchable text&lt;/strong&gt;, for example to archive scans and find them later.&lt;/li&gt;
&lt;li&gt;The volume is high and the layout never changes.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What AI document processing does
&lt;/h2&gt;

&lt;p&gt;AI document processing (often called intelligent document processing, or IDP) starts where OCR stops. It reads the text the way a person would and returns the &lt;strong&gt;fields you ask for&lt;/strong&gt;, already structured: supplier name, invoice number, date, line items, total.&lt;/p&gt;

&lt;p&gt;Because it understands the content rather than the position, it:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Handles layouts it hasn't seen.&lt;/strong&gt; A new supplier's invoice works without building a template.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Reads tables and line items&lt;/strong&gt; as tables, including ones that run over several pages.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Gives a confidence level for each field&lt;/strong&gt;, so uncertain values can go to a person.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Works on more than forms&lt;/strong&gt;: resumes, contracts, certificates of insurance, receipts, patient forms and emails, where the information isn't in fixed boxes.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Side by side
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;&lt;/th&gt;
&lt;th&gt;Traditional OCR&lt;/th&gt;
&lt;th&gt;AI document processing&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;What it does&lt;/td&gt;
&lt;td&gt;Reads text from an image&lt;/td&gt;
&lt;td&gt;Understands the document and returns the fields you need&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;New layouts&lt;/td&gt;
&lt;td&gt;Needs a template per layout&lt;/td&gt;
&lt;td&gt;Handles new layouts without setup&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Tables and line items&lt;/td&gt;
&lt;td&gt;Often breaks&lt;/td&gt;
&lt;td&gt;Reads them as tables&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Checking&lt;/td&gt;
&lt;td&gt;None&lt;/td&gt;
&lt;td&gt;Validation rules + confidence per field&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Output&lt;/td&gt;
&lt;td&gt;Text&lt;/td&gt;
&lt;td&gt;Structured data for your system&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  When plain OCR is still enough
&lt;/h2&gt;

&lt;p&gt;AI isn't always the answer. Stick with OCR (or the OCR built into your scanner or document management system) if:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;You already have a legacy system using OCR that doesn't need new layouts or workflow changes.&lt;/li&gt;
&lt;li&gt;You only need the documents to be &lt;strong&gt;searchable&lt;/strong&gt;, not their data in another system.&lt;/li&gt;
&lt;li&gt;You process &lt;strong&gt;one fixed form&lt;/strong&gt; and the template already works.&lt;/li&gt;
&lt;li&gt;Your documents are &lt;strong&gt;clean, typed and identical&lt;/strong&gt; in layout, and nobody is spending time fixing errors.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If that describes your process, an AI project would add cost without much benefit.&lt;/p&gt;

&lt;h2&gt;
  
  
  When AI document processing pays off
&lt;/h2&gt;

&lt;p&gt;It's worth looking at AI when:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Layouts vary.&lt;/strong&gt; Invoices from dozens of suppliers, resumes in every format, purchase orders from different customers.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Someone is still re-typing.&lt;/strong&gt; If people copy values from PDFs into your ERP, CRM or applicant tracking system, that's the time AI saves.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Templates keep breaking.&lt;/strong&gt; Every new supplier or form version means a new template and a support call.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You need line items&lt;/strong&gt;, not just header fields.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The information isn't in boxes.&lt;/strong&gt; Contracts, resumes and emails don't have fixed fields at all.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How it fits into your process
&lt;/h2&gt;

&lt;p&gt;Whichever technology reads the document, the process around it matters as much:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Documents come in&lt;/strong&gt; by email, upload, scanner or a shared folder.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The AI extracts the fields&lt;/strong&gt; you defined for that document type.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Validation rules check them&lt;/strong&gt;: required fields present, totals add up, the supplier or customer exists in your system.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Confident results go straight through. Anything uncertain goes to a review queue&lt;/strong&gt;, where a person sees the document next to the extracted fields and corrects them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The approved data goes into your system&lt;/strong&gt; through its API, a database import or a file it already reads.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;The review step is what makes this safe. You decide which documents can go straight into your ERP or other systems and which need a person to check them. As the results on your documents prove reliable, you can loosen the checks.&lt;/p&gt;

&lt;h2&gt;
  
  
  What about handwriting and poor scans?
&lt;/h2&gt;

&lt;p&gt;Both OCR and AI do worse on handwriting, faded copies and phone photos taken at an angle. AI copes better because it can use context (a value next to "Due date" is probably a date), but it isn't magic. That's why the review queue matters, and why we always measure accuracy on a sample of &lt;strong&gt;your own&lt;/strong&gt; documents during a pilot instead of quoting a general number.&lt;/p&gt;

&lt;h2&gt;
  
  
  What we use
&lt;/h2&gt;

&lt;p&gt;We built our own platform for this, &lt;a href="https://dquest.ai" rel="noopener noreferrer"&gt;dQuest.ai&lt;/a&gt;. It processes any document type, lets you define new document types, lets you choose the AI model (Azure OpenAI, OpenAI or Google Gemini, on your account or ours) and where documents are stored (SharePoint, Google Drive, Dropbox and others), and includes the review step. Because the platform is ours, we can add a document type, a rule or an integration your process needs.&lt;/p&gt;

&lt;p&gt;For example, a staffing company used it to read candidate resumes and feed the candidate data into its applicant tracking system. It was running within a week and handled 250 to 400 resumes a week.&lt;/p&gt;

&lt;p&gt;You can try standard documents on dQuest.ai's free plan. For healthcare documents, we sign a HIPAA business associate agreement.&lt;/p&gt;

&lt;h2&gt;
  
  
  Frequently asked questions
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Is AI document processing the same as OCR?
&lt;/h3&gt;

&lt;p&gt;No. OCR turns an image into text. AI document processing understands that text and returns structured fields, such as the invoice total or the candidate's email address, even when layouts differ.&lt;/p&gt;

&lt;h3&gt;
  
  
  Does AI document processing still use OCR?
&lt;/h3&gt;

&lt;p&gt;Often, yes. For scans and photos, text recognition is usually the first step, and the AI works on what it produces. The difference is what happens next.&lt;/p&gt;

&lt;h3&gt;
  
  
  Do I need a template for each supplier?
&lt;/h3&gt;

&lt;p&gt;Not with AI document processing. It doesn't rely on the position of fields, so new layouts work without setup.&lt;/p&gt;

&lt;h3&gt;
  
  
  How accurate is AI document processing?
&lt;/h3&gt;

&lt;p&gt;It depends on the document type and scan quality. Clean, typed documents extract very reliably; handwriting and poor scans less so. The best measure is a pilot on a sample of your own documents, with uncertain fields sent to a person.&lt;/p&gt;

&lt;h3&gt;
  
  
  Can the data go straight into my ERP?
&lt;/h3&gt;

&lt;p&gt;Yes. The extracted data can be posted into your ERP, CRM or line-of-business system through its API, or through a database import or file it already reads. See &lt;a href="https://www.sharpquest.com/ai-integration/ai-erp-integration/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-ai-doc-ocr" rel="noopener noreferrer"&gt;AI ERP integration&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;SharpQuest builds custom web, mobile and AI software for businesses across the US. More on how we set up &lt;a href="https://www.sharpquest.com/ai-integration/ai-document-processing/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-ai-doc-ocr" rel="noopener noreferrer"&gt;AI document processing&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>ocr</category>
      <category>automation</category>
      <category>productivity</category>
    </item>
    <item>
      <title>What Is a User Story? User Story vs Task, Feature, Epic and Bug</title>
      <dc:creator>Piyush Bhatt</dc:creator>
      <pubDate>Mon, 09 Nov 2026 02:13:44 +0000</pubDate>
      <link>https://dev.to/piyushpbhatt/what-is-a-user-story-user-story-vs-task-feature-epic-and-bug-81f</link>
      <guid>https://dev.to/piyushpbhatt/what-is-a-user-story-user-story-vs-task-feature-epic-and-bug-81f</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://www.sharpquest.com/blog/posts/2020/december/what-is-what-is-not-a-user-story-or-a-backlog-item/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-user-story" rel="noopener noreferrer"&gt;SharpQuest blog&lt;/a&gt;, written by Piyush Bhatt.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key takeaways&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;A user story, or product backlog item (PBI), describes what a user or the system should be able to do and why, not how it will be built.&lt;/li&gt;
&lt;li&gt;Write it as "As a [user], I want [capability], so that [benefit]", and add acceptance criteria so everyone agrees when it’s done.&lt;/li&gt;
&lt;li&gt;A story isn’t a task (how the work gets done), a feature or epic (a group of stories), or always one developer’s work. Developers add their own tasks and estimates under it.&lt;/li&gt;
&lt;li&gt;Some backlog items are system requirements users never see, such as purging history older than three months. They still belong in the backlog.&lt;/li&gt;
&lt;li&gt;A bug is undesired behavior, but it can be planned and tracked the same way as a story.&lt;/li&gt;
&lt;li&gt;AI assistants can draft stories and acceptance criteria from meeting notes, but the product owner still has to check them.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;A user story describes something a user should be able to do, and why it matters to them. It says &lt;em&gt;what&lt;/em&gt; the software should do, not &lt;em&gt;how&lt;/em&gt; the team will build it. One of the most common Agile mistakes is mixing the two: developers write implementation tasks and call them user stories, and product owners write feature requests so broad that nobody can estimate them. Knowing the difference helps teams plan, estimate, prioritize and deliver better.&lt;/p&gt;

&lt;p&gt;This is worth thinking about if you’re a:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;software development manager&lt;/li&gt;
&lt;li&gt;product owner or product manager&lt;/li&gt;
&lt;li&gt;project manager or Scrum Master&lt;/li&gt;
&lt;li&gt;business analyst&lt;/li&gt;
&lt;li&gt;quality analyst&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;and especially when you manage a team that includes some or all of these roles.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why user stories matter
&lt;/h2&gt;

&lt;p&gt;If you’re a founder, an operations manager or an IT director paying for software, user stories aren’t just Agile vocabulary. Well-written stories reduce misunderstandings, make estimates more reliable, and keep the team focused on business value rather than technical detail. A project that starts with clear stories and acceptance criteria leaves less room for change requests, budget surprises and missed expectations than one that starts with a loose list of features.&lt;/p&gt;

&lt;p&gt;This is why discovery and requirements workshops are a standard part of our process for larger software projects. Investing time in defining workflows, user stories, and acceptance criteria before development begins helps teams align expectations and identify gaps before they become expensive changes later in the project.&lt;/p&gt;

&lt;p&gt;When we started out as a &lt;a href="https://www.sharpquest.com/software-development-company/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-user-story" rel="noopener noreferrer"&gt;software development firm&lt;/a&gt;, there was no distinction between the tasks developers wanted to do and the things clients were asking for. It was all one pool, spread across Excel files, Google Sheets and emails. That doesn’t mean we weren’t efficient: a small team on a small project can be very efficient that way. But when the projects and teams get bigger, you need the right tools and a better process, and that’s when you start asking what is &lt;em&gt;not&lt;/em&gt; a user story.&lt;/p&gt;

&lt;p&gt;In Azure DevOps and Jira, user stories are commonly managed as &lt;strong&gt;product backlog items (PBIs)&lt;/strong&gt;. Before Agile tools, the same thing was usually called a requirement, or a user requirement.&lt;/p&gt;

&lt;h2&gt;
  
  
  The user story format
&lt;/h2&gt;

&lt;p&gt;Most teams write the story in three parts:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;As a&lt;/strong&gt; [type of user], &lt;strong&gt;I want&lt;/strong&gt; [some capability], &lt;strong&gt;so that&lt;/strong&gt; [some benefit].&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;For example:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;As an administrator, I want to search users by email, so that I can quickly find a customer’s account.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;In our own backlogs, the PBI title is a short version a non-technical reader can follow, and the "As a" sentence goes in the description:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Title:&lt;/strong&gt; Website visitor should be able to submit the feedback form&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Description:&lt;/strong&gt; As a website visitor, I want to submit feedback, so that the site owners can consider my opinion in future updates.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The "so that" part is the one teams skip most often, and it’s the most useful. It tells the developer why the story exists, which helps them make the small decisions the story doesn’t spell out.&lt;/p&gt;

&lt;h2&gt;
  
  
  User story vs task, feature, epic and bug
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Item&lt;/th&gt;
&lt;th&gt;What it describes&lt;/th&gt;
&lt;th&gt;Example&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Epic&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;A large body of work, usually several features, delivered over many sprints&lt;/td&gt;
&lt;td&gt;Customer self-service portal&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Feature&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;A group of related stories&lt;/td&gt;
&lt;td&gt;User account management&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;User story / PBI&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;What one user (or the system) should be able to do, and why&lt;/td&gt;
&lt;td&gt;User should be able to reset their password&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Task&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;How the team gets part of a story done; hours, not days&lt;/td&gt;
&lt;td&gt;Create the password reset API endpoint&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Bug&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Behavior that’s wrong and needs correcting&lt;/td&gt;
&lt;td&gt;Password reset email isn’t sent&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  What a user story is not
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;A user story is not a task.&lt;/strong&gt; If you’re a developer, or have a strong technical background, this is the one to remember. The work we do as developers, such as creating database tables, taking backups, adding a login form, securing an API or validating an email address, is not a user story. Developers naturally think in terms of the tasks they’ll perform, and it’s easy to enter those in Azure DevOps or Jira as stories. They’re tasks.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A user story is not &lt;em&gt;how&lt;/em&gt;.&lt;/strong&gt; A backlog item shouldn’t describe how a feature will be implemented. Whether you add a drop-down, a new button or a new API is a detail of how you’ll do it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A user story is not a feature.&lt;/strong&gt; A story doesn’t need to be as small as a task, but it shouldn’t be as broad as a whole feature either. "Search" is a feature. "User should be able to search products by keyword" is a story.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A user story is not always one programmer’s work.&lt;/strong&gt; In a small team, especially one with a developer who can do a bit of everything, you can assign a story to one person. In a larger team with specialists, one story may need a front-end developer, a database developer and an API developer. Then the story stays with one owner who’s accountable for it (see below), and each developer owns their tasks.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The BA or manager doesn’t create the tasks.&lt;/strong&gt; Tasks describe how a story will be built, and that’s the development team’s job. The developers working on the story add their tasks and estimates under it.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What a user story is
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;A user story is &lt;em&gt;what&lt;/em&gt;, not &lt;em&gt;how&lt;/em&gt;.&lt;/strong&gt; When you read it, it should sound like what the user needs, not how it will be built. Examples:

&lt;ul&gt;
&lt;li&gt;User should be able to log in with email and password.&lt;/li&gt;
&lt;li&gt;Admin user should be able to see a list of all application users.&lt;/li&gt;
&lt;li&gt;Admin user should be able to define roles from a set of predefined permissions.&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Most user stories start with the user.&lt;/strong&gt; Notice that each example starts with a user, and that means the end user. What the user should be able to do is the focus, not what the developer should do. Some backlog items, though, are system requirements that matter to the business even though users never see them directly.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A product backlog item can also be a system requirement.&lt;/strong&gt; For example:

&lt;ul&gt;
&lt;li&gt;System should purge user history records older than three months, so that the database doesn’t grow without limit.&lt;/li&gt;
&lt;li&gt;System should keep a log of user logins for the past three months, so that administrators can audit access.&lt;/li&gt;
&lt;li&gt;System should archive invoices older than seven years, so that regulatory and retention requirements are met.&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;These are the client’s requirements, even though end users care little about them. That’s why many teams call backlog items PBIs rather than user stories: "PBI" covers both.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;A story has one accountable owner.&lt;/strong&gt; Depending on the team, that may be the product owner, a business analyst, the Scrum Master, a tech lead or a developer. In small teams without a full-time BA or lead, it’s normal for the developer doing the work to own the story. What matters is that one person is accountable for clarifying the requirement and for delivering it to the stakeholders.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Developers create their tasks for each story.&lt;/strong&gt; The developers add tasks for each story, with an estimate for each. In our teams a task is typically 1-4 hours and has to finish inside the sprint. It’s fine to start with a single analysis task for the effort and split it further once the work is better understood.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  User stories need acceptance criteria
&lt;/h2&gt;

&lt;p&gt;A story says what the user wants. Acceptance criteria say how everyone will know it’s done. Without them, the developer, the tester and the product owner can each have a different picture of "finished".&lt;/p&gt;

&lt;p&gt;Many teams write acceptance criteria in a Given/When/Then form. Here’s the feedback-form story from above:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Given&lt;/strong&gt; I’m a guest or signed-in user&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;When&lt;/strong&gt; I open the Feedback page&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Then&lt;/strong&gt; the system shows the Submit Feedback form with required Email, Name and Comment fields&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;When&lt;/strong&gt; I fill in all three fields with valid values and click Submit Feedback&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Then&lt;/strong&gt; the system saves my feedback, shows "You’ve successfully submitted your feedback", and clears the form&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;A simple checklist works too. For "User should be able to reset their password":&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;The reset link is emailed to the registered address.&lt;/li&gt;
&lt;li&gt;The link expires after 24 hours.&lt;/li&gt;
&lt;li&gt;The user can set a new password that meets the password rules.&lt;/li&gt;
&lt;li&gt;The user sees a confirmation message, and can sign in with the new password.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;In our process, the product owner adds acceptance criteria before the backlog refinement meeting, the development team checks they’re clear before committing to the story, and the developer verifies every criterion before opening a pull request.&lt;/p&gt;

&lt;h2&gt;
  
  
  What makes a good user story?
&lt;/h2&gt;

&lt;p&gt;A common checklist is &lt;strong&gt;INVEST&lt;/strong&gt;, from the &lt;a href="https://agilealliance.org/glossary/invest/" rel="noopener noreferrer"&gt;Agile Alliance&lt;/a&gt;. A good story is:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Independent:&lt;/strong&gt; it can be built and released without waiting for another story.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Negotiable:&lt;/strong&gt; it states the need, and leaves room to agree on the details.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Valuable:&lt;/strong&gt; a user or the business gets something from it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Estimable:&lt;/strong&gt; the team understands it well enough to estimate it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Small:&lt;/strong&gt; it fits comfortably in one sprint.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Testable:&lt;/strong&gt; it has acceptance criteria someone can check.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Common user story mistakes
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Mistake&lt;/th&gt;
&lt;th&gt;Weak story&lt;/th&gt;
&lt;th&gt;Better approach&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Writing technical tasks as stories&lt;/td&gt;
&lt;td&gt;Create the users table&lt;/td&gt;
&lt;td&gt;Focus on the user outcome: "Admin user should be able to add a new application user"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Describing how instead of what&lt;/td&gt;
&lt;td&gt;Add a drop-down for country&lt;/td&gt;
&lt;td&gt;State the need: "User should be able to choose their country when registering, so that prices show in their currency"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Stories too large for one sprint&lt;/td&gt;
&lt;td&gt;User management&lt;/td&gt;
&lt;td&gt;Split into smaller stories: "Admin user should be able to deactivate a user, so that former staff can’t sign in"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;No user benefit&lt;/td&gt;
&lt;td&gt;User should be able to export reports&lt;/td&gt;
&lt;td&gt;Add the "so that": "…so that the finance team can reconcile invoices in Excel"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Missing or untestable acceptance criteria&lt;/td&gt;
&lt;td&gt;Make the search faster&lt;/td&gt;
&lt;td&gt;Define when it’s done: "Product search returns results within two seconds for the full catalog"&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  An example: from feature to tasks
&lt;/h2&gt;

&lt;p&gt;Here’s how the levels fit together on a typical custom software project, such as a customer service portal:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4fe6toni5l0g8skfo76c.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4fe6toni5l0g8skfo76c.png" alt="Diagram: an epic contains features, a feature contains user stories, each story has acceptance criteria, and developers break each story into tasks" width="800" height="520"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Epic:&lt;/strong&gt; Customer service portal&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Feature:&lt;/strong&gt; Customer search&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;User story:&lt;/strong&gt; As a customer service representative, I want to search customers by name or email, so that I can find their account while they’re on the phone.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Acceptance criteria:&lt;/strong&gt; partial names match; results show name, email and account number; an email search returns an exact match first.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Developer tasks:&lt;/strong&gt;

&lt;ul&gt;
&lt;li&gt;Add a full-text index on the customer name and email columns&lt;/li&gt;
&lt;li&gt;Build the search API endpoint&lt;/li&gt;
&lt;li&gt;Build the search screen&lt;/li&gt;
&lt;li&gt;Write automated tests for the endpoint and the screen&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The product owner wrote one story. The developers turned it into four tasks, estimated each, and owned them. Nobody had to write "build the search API" as a user story.&lt;/p&gt;

&lt;p&gt;We run every project this way, with stories, acceptance criteria and sprints; see &lt;a href="https://www.sharpquest.com/engagement-model/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-user-story" rel="noopener noreferrer"&gt;how we work&lt;/a&gt;. The &lt;a href="https://www.sharpquest.com/blog/posts/2020/december/software-development-processes-daily-stand-up/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-user-story" rel="noopener noreferrer"&gt;daily stand-up&lt;/a&gt; is where the team tracks those tasks day to day.&lt;/p&gt;

&lt;h2&gt;
  
  
  Story and bug
&lt;/h2&gt;

&lt;p&gt;A story is a desired behavior that a system or application should have. It brings some benefit to the user or the system.&lt;/p&gt;

&lt;p&gt;A bug is an undesired behavior the system has and that needs correcting. It gets in the way of the benefit a story brings.&lt;/p&gt;

&lt;p&gt;The two are different, but you can manage a bug in your project tool the same way as a story: it has an owner, it may need several team members, and it can have tasks done by different people.&lt;/p&gt;

&lt;h2&gt;
  
  
  Frequently asked questions
&lt;/h2&gt;

&lt;h3&gt;
  
  
  What is the difference between a user story and a task?
&lt;/h3&gt;

&lt;p&gt;A user story describes what a user needs and why. A task describes a piece of work the team does to deliver that story, such as building an API endpoint or writing tests. A story is written by or for the product owner; tasks are written by the developers.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is a product backlog item the same as a user story?
&lt;/h3&gt;

&lt;p&gt;Mostly. In Azure DevOps and Jira, "product backlog item" is the general name for an item in the backlog. It covers user stories and also system requirements that no user sees directly, such as purging old records.&lt;/p&gt;

&lt;h3&gt;
  
  
  Who writes user stories?
&lt;/h3&gt;

&lt;p&gt;Usually the product owner or a business analyst, often with the client. Developers and testers help refine them, but they shouldn’t turn their own tasks into stories.&lt;/p&gt;

&lt;h3&gt;
  
  
  How big should a user story be?
&lt;/h3&gt;

&lt;p&gt;Small enough to finish in one sprint, and big enough that a user gets something from it. If a story needs more than one sprint, split it; if it only describes one technical step, it’s a task.&lt;/p&gt;

&lt;h3&gt;
  
  
  What are acceptance criteria?
&lt;/h3&gt;

&lt;p&gt;The conditions a story must meet to be accepted as done. They’re often written as Given/When/Then scenarios or a short checklist, and the developer and tester check them before the story is closed.&lt;/p&gt;

&lt;h3&gt;
  
  
  Should technical work be written as a user story?
&lt;/h3&gt;

&lt;p&gt;Technical work that delivers something the client needs, such as audit logging or data retention, can be a backlog item written as "System should … so that …". Technical steps that only serve a story, such as creating a table, are tasks under that story.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is a bug a user story?
&lt;/h3&gt;

&lt;p&gt;No. A story adds behavior the system should have; a bug corrects behavior it shouldn’t have. You can still track and plan a bug the same way as a story.&lt;/p&gt;

&lt;h3&gt;
  
  
  Can AI write user stories?
&lt;/h3&gt;

&lt;p&gt;AI assistants can turn meeting notes or a requirements document into draft stories and acceptance criteria quickly, and they’re good at spotting missing edge cases. But they don’t know your users or your business rules. Tools such as Microsoft Copilot, ChatGPT, Claude and Gemini are useful for drafting backlog items, but treat them as writing assistants, not product owners: the product owner still has to review every story before it goes into a sprint.&lt;/p&gt;




&lt;p&gt;User stories focus on the value a user or the system gets, while tasks describe how developers deliver that value. Teams that keep stories separate from implementation details have clearer backlogs, better estimates, and better communication between business stakeholders and developers. At SharpQuest, we write user stories and acceptance criteria during discovery and refine them throughout development, so clients know what’s being built, why, and how completion will be measured.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;SharpQuest builds custom web, mobile and AI software for businesses across the US. More on how we run projects in &lt;a href="https://www.sharpquest.com/engagement-model/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-user-story" rel="noopener noreferrer"&gt;how we work&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>agile</category>
      <category>scrum</category>
      <category>productivity</category>
      <category>projectmanagement</category>
    </item>
    <item>
      <title>How to Migrate a Xamarin.Forms App to .NET MAUI 10 (2026 Guide)</title>
      <dc:creator>Piyush Bhatt</dc:creator>
      <pubDate>Wed, 04 Nov 2026 21:00:37 +0000</pubDate>
      <link>https://dev.to/piyushpbhatt/how-to-migrate-a-xamarinforms-app-to-net-maui-10-2026-guide-3dk3</link>
      <guid>https://dev.to/piyushpbhatt/how-to-migrate-a-xamarinforms-app-to-net-maui-10-2026-guide-3dk3</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://www.sharpquest.com/blog/posts/2024/march/how-to-migrate-mobile-apps-from-xamarin-and-xamarin-forms-to-net-maui/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-xamarin-maui" rel="noopener noreferrer"&gt;SharpQuest blog&lt;/a&gt;, written by Greg Kabacinski and Rutul Mehta from our .NET MAUI team. First published in March 2024 and updated for .NET 10, current tooling and the 2026 app store requirements.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key takeaways&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Microsoft ended support for Xamarin on May 1, 2024. Xamarin's last versions no longer meet Google Play and App Store requirements, so Xamarin apps can't ship updates.&lt;/li&gt;
&lt;li&gt;.NET MAUI is the successor to Xamarin.Forms: one project and one codebase for Android, iOS, macOS and Windows. Migrate to .NET 10 (MAUI 10), the current long-term support release.&lt;/li&gt;
&lt;li&gt;Before migrating, update the app to Xamarin.Forms 5 and current dependencies.&lt;/li&gt;
&lt;li&gt;There are two ways to migrate: a migration tool, or a new .NET MAUI project with your code copied in. We combine both. The .NET Upgrade Assistant is deprecated; the GitHub Copilot modernization agent in Visual Studio replaces it.&lt;/li&gt;
&lt;li&gt;Plan time for layouts, custom renderers and third-party packages. These take the most effort.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;.NET Multi-platform App UI (.NET MAUI) is the successor to Xamarin.Forms and Microsoft's framework for cross-platform apps with .NET. It lets you write code once and run it on Android, iOS, macOS and Windows. .NET MAUI is based on Xamarin.Forms, but it brings many improvements:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;A single project instead of one project per platform.&lt;/li&gt;
&lt;li&gt;A unified app model for creating and configuring apps across platforms.&lt;/li&gt;
&lt;li&gt;A modernized control library built on native controls, with support for customization.&lt;/li&gt;
&lt;li&gt;A new handler architecture that improves performance and reduces app size.&lt;/li&gt;
&lt;li&gt;Community toolkits with additional controls, behaviors and utilities.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why Xamarin apps need to move
&lt;/h2&gt;

&lt;p&gt;Microsoft ended support for all Xamarin SDKs, including Xamarin.Forms, on May 1, 2024. There are no more bug fixes, security updates or new features. The last Xamarin releases target Android API 34 and Xcode 15.&lt;/p&gt;

&lt;p&gt;That now blocks releases. Since April 28, 2026, Apple requires App Store uploads to be built with Xcode 26, and since August 31, 2026, Google Play requires new apps and updates to target Android 16 (API 36). A Xamarin app already in the stores keeps running for now, but you can't publish an update to it, not even a security fix.&lt;/p&gt;

&lt;p&gt;If you're still on Xamarin, plan the migration to .NET MAUI now. The rest of this guide covers the steps, and the problems you're likely to hit along the way. (If you're still deciding whether to migrate or rewrite, see &lt;a href="https://www.sharpquest.com/blog/posts/2026/october/still-on-xamarin-in-2026/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-xamarin-maui" rel="noopener noreferrer"&gt;Still on Xamarin in 2026?&lt;/a&gt;)&lt;/p&gt;

&lt;h2&gt;
  
  
  Migration benefits
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Current .NET and C#.&lt;/strong&gt; You get the latest C# language features, .NET Hot Reload and the performance work in each .NET release. Target .NET 10 today; see &lt;a href="https://devblogs.microsoft.com/dotnet/performance-improvements-in-net-10/" rel="noopener noreferrer"&gt;Performance improvements in .NET 10&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;One project, one codebase&lt;/strong&gt; for all platforms, which simplifies development and maintenance.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Native controls and handlers&lt;/strong&gt; you can customize, plus the community toolkits.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A future-proof app&lt;/strong&gt; that runs on the latest versions of Android and iOS, with ongoing updates from Microsoft and the .NET community.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Stay current after you migrate, too. Each major version of .NET MAUI is supported for 6 months after the next one ships: MAUI 8 ended in May 2025 and MAUI 9 in May 2026. Plan a yearly upgrade, just as you would for any other .NET app.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prerequisites
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Visual Studio 2026&lt;/strong&gt; (or Visual Studio 2022 version 17.14 or later) with the &lt;strong&gt;.NET Multi-platform App UI development&lt;/strong&gt; workload. On macOS, use Visual Studio Code with the .NET MAUI extension; Visual Studio for Mac has been retired.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A Mac with Xcode 26&lt;/strong&gt; to build and sign the iOS app; the App Store requires it since April 28, 2026.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;.NET 10 SDK.&lt;/strong&gt; Target .NET 10 (MAUI 10). .NET 8 and .NET 9 reach end of support on November 10, 2026.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Xamarin.Forms 5.0&lt;/strong&gt; in your app. Xamarin.Forms 4.8 is the minimum, but upgrading to 5.0 first removes many differences before you start. Shared code should target .NET Standard 2.0 or later.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Up-to-date dependencies.&lt;/strong&gt; Update your NuGet packages, especially those that have .NET-compatible versions.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Migration options
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;1. Tool-assisted upgrade.&lt;/strong&gt; The .NET Upgrade Assistant is now deprecated. Microsoft's replacement is the &lt;a href="https://learn.microsoft.com/en-us/dotnet/core/porting/github-copilot-app-modernization/faq" rel="noopener noreferrer"&gt;GitHub Copilot modernization agent&lt;/a&gt;, included in Visual Studio 2026 (and Visual Studio 2022 17.14.17 and later). It analyzes your projects and dependencies, proposes a plan, applies code changes and commits each step so you can review it or roll it back. Treat its output as a first pass: review every change, and expect to finish layouts, renderers and packages by hand.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. New .NET MAUI project.&lt;/strong&gt; Create a new .NET MAUI project with the same name as your Xamarin.Forms app, then copy in your code files, XAML files, images, fonts and other assets, and update the code to the .NET MAUI APIs. This gives you the most control and a clean single-project structure, but it takes more effort. Microsoft documents the process in &lt;a href="https://learn.microsoft.com/en-us/dotnet/maui/migration/multi-project-to-multi-project" rel="noopener noreferrer"&gt;Manually upgrade a Xamarin.Forms app&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Our approach combines both.&lt;/strong&gt; We run the tool first to get the project files, namespaces and package references converted, then create a new single-project MAUI app and move the code across, using the converted project as a reference. We've moved a &lt;a href="https://www.sharpquest.com/portfolio/home-care-and-hospice-patient-caregiver-app/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-xamarin-maui" rel="noopener noreferrer"&gt;home-care patient app&lt;/a&gt; and an &lt;a href="https://www.sharpquest.com/portfolio/employee-time-tracking-mobile-app/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-xamarin-maui" rel="noopener noreferrer"&gt;employee time-tracking app&lt;/a&gt; from Xamarin to .NET MAUI this way.&lt;/p&gt;

&lt;h2&gt;
  
  
  Migration steps
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;1. Back up and branch.&lt;/strong&gt; Make sure the Xamarin.Forms app is in source control (such as Git) so you can return to a working state at any point.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. Update to Xamarin.Forms 5.&lt;/strong&gt; Update the app and its dependencies to the latest versions and make sure it still runs. This keeps the API differences with .NET MAUI small and moves you onto .NET-compatible package versions where they exist.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. Convert the project.&lt;/strong&gt; Use the tool-assisted upgrade, a new MAUI project, or both, as described above.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;4. Fix namespaces.&lt;/strong&gt; Compare the files of the Xamarin.Forms app with the converted MAUI project and make sure the namespaces match Microsoft's migration guidance.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;5. Move startup code to &lt;code&gt;MauiProgram&lt;/code&gt;.&lt;/strong&gt; .NET MAUI has a platform-independent startup class, &lt;code&gt;MauiProgram&lt;/code&gt;, as the app's entry point. If you've used ASP.NET Core, it will look familiar: an app builder where you register services in the dependency injection container and configure the app. Packages built for MAUI are usually set up in &lt;code&gt;MauiProgram&lt;/code&gt; rather than in the platform startup classes (&lt;code&gt;MainApplication&lt;/code&gt; and &lt;code&gt;MainActivity&lt;/code&gt; on Android, &lt;code&gt;AppDelegate&lt;/code&gt; on iOS). Some configuration, such as push notifications, still belongs in those classes. &lt;code&gt;MainActivity&lt;/code&gt; and &lt;code&gt;AppDelegate&lt;/code&gt; also look quite different from Xamarin's, so compare yours with a new project created from the MAUI template.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;6. Update APIs.&lt;/strong&gt; Most of Xamarin.Essentials is now built into .NET MAUI (see &lt;a href="https://learn.microsoft.com/en-us/dotnet/maui/platform-integration/" rel="noopener noreferrer"&gt;platform integration&lt;/a&gt;) and is included automatically with &lt;code&gt;&amp;lt;UseMaui&amp;gt;true&amp;lt;/UseMaui&amp;gt;&lt;/code&gt; in your project file. To find where an API moved, use the &lt;a href="https://learn.microsoft.com/en-us/dotnet/api/?view=net-maui-10.0" rel="noopener noreferrer"&gt;.NET API browser&lt;/a&gt;. The Xamarin Community Toolkit is now &lt;code&gt;CommunityToolkit.Maui&lt;/code&gt;, with many of the same features. If you use &lt;code&gt;MessagingCenter&lt;/code&gt;, replace it with &lt;code&gt;WeakReferenceMessenger&lt;/code&gt; from &lt;code&gt;CommunityToolkit.Mvvm&lt;/code&gt;. For lists, &lt;code&gt;CollectionView&lt;/code&gt; is the recommended replacement for &lt;code&gt;ListView&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;7. Fix layouts.&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;RelativeLayout&lt;/code&gt; is removed. Replace it with a &lt;code&gt;Grid&lt;/code&gt; or another layout. (The compatibility namespace still lets you use it, but we don't recommend it.)&lt;/li&gt;
&lt;li&gt;The &lt;code&gt;...AndExpand&lt;/code&gt; options on &lt;code&gt;StackLayout&lt;/code&gt; are deprecated. Outside a &lt;code&gt;StackLayout&lt;/code&gt; they act as if &lt;code&gt;AndExpand&lt;/code&gt; weren't there (&lt;code&gt;FillAndExpand&lt;/code&gt; behaves like &lt;code&gt;Fill&lt;/code&gt;). Replace them with a &lt;code&gt;Grid&lt;/code&gt;, or use the newer &lt;code&gt;HorizontalStackLayout&lt;/code&gt; and &lt;code&gt;VerticalStackLayout&lt;/code&gt;, which behave like &lt;code&gt;StackLayout&lt;/code&gt; with &lt;code&gt;Orientation="Horizontal"&lt;/code&gt; or &lt;code&gt;"Vertical"&lt;/code&gt;, with a few differences.&lt;/li&gt;
&lt;li&gt;Even without those layouts, the app will look slightly different: margins, padding and the size of elements and text change. You can restore some of Xamarin.Forms' defaults as described in &lt;a href="https://learn.microsoft.com/en-us/dotnet/maui/migration/layouts#default-layout-value-changes-from-xamarinforms" rel="noopener noreferrer"&gt;Layout behavior changes from Xamarin.Forms&lt;/a&gt;, but expect to adjust screens by hand to make them look as they did.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;8. Replace custom renderers with handlers.&lt;/strong&gt; If a renderer derives from &lt;code&gt;FrameRenderer&lt;/code&gt;, &lt;code&gt;ListViewRenderer&lt;/code&gt;, &lt;code&gt;ShellRenderer&lt;/code&gt;, &lt;code&gt;TableViewRenderer&lt;/code&gt; or &lt;code&gt;VisualElementRenderer&lt;/code&gt;, you can derive it from the matching class in &lt;code&gt;Microsoft.Maui.Controls.Compatibility&lt;/code&gt; to get the app compiling quickly (&lt;a href="https://learn.microsoft.com/en-us/dotnet/maui/migration/custom-renderers" rel="noopener noreferrer"&gt;Reuse custom renderers&lt;/a&gt;). The long-term fix is to move each renderer to a handler (&lt;a href="https://learn.microsoft.com/en-us/dotnet/maui/migration/renderer-to-handler" rel="noopener noreferrer"&gt;Migrate a custom renderer to a handler&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;9. Find compatible packages.&lt;/strong&gt; Getting every library to work in MAUI is often the hardest part, and harder still when you upgrade package versions and the .NET version at the same time. It looks different for every app: some packages have MAUI versions, some have replacements, and for some you'll need to change the code or the source yourself. Our solutions for Firebase and Twilio Video are below.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;10. Test on real devices and publish.&lt;/strong&gt; Test on Android and iOS devices (not just emulators), fix layout and behavior differences, and check the app against your release checklist. Build with the SDKs the stores require today: Xcode 26 for the App Store, and Android 16 (API 36) as the target for Google Play.&lt;/p&gt;

&lt;h2&gt;
  
  
  Package migration: Firebase and Twilio Video
&lt;/h2&gt;

&lt;p&gt;These notes come from our migration of a home-care patient app that uses Firebase, and Twilio Video for its video visits.&lt;/p&gt;

&lt;h3&gt;
  
  
  Firebase
&lt;/h3&gt;

&lt;p&gt;You have two options:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Keep Microsoft's Firebase binding packages.&lt;/strong&gt; They're the same packages Xamarin used, so your project will still reference packages whose names start with &lt;code&gt;Xamarin.&lt;/code&gt; after the migration.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Switch to &lt;a href="https://github.com/TobiasBuchholz/Plugin.Firebase" rel="noopener noreferrer"&gt;Plugin.Firebase&lt;/a&gt;&lt;/strong&gt;, a wrapper built for MAUI. It takes more code changes, but the &lt;a href="https://github.com/TobiasBuchholz/Plugin.Firebase/tree/development/docs" rel="noopener noreferrer"&gt;setup is well documented&lt;/a&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;With either package, you may run into this build error: &lt;strong&gt;"Could not find a part of the path."&lt;/strong&gt; Some paths in the MAUI build are too long for Windows. In our case &lt;code&gt;Xamarin.Firebase.iOS.CloudMessaging&lt;/code&gt; failed to build. To fix it, enable long paths and move the NuGet package folder to a short path:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Close all Visual Studio instances.&lt;/li&gt;
&lt;li&gt;Run PowerShell as Administrator and enable long paths:
&lt;/li&gt;
&lt;/ol&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight powershell"&gt;&lt;code&gt;&lt;span class="w"&gt;   &lt;/span&gt;&lt;span class="n"&gt;New-ItemProperty&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nt"&gt;-Path&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"HKLM:\SYSTEM\CurrentControlSet\Control\FileSystem"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nt"&gt;-Name&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"LongPathsEnabled"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nt"&gt;-Value&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nx"&gt;1&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nt"&gt;-PropertyType&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nx"&gt;DWORD&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nt"&gt;-Force&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;ol&gt;
&lt;li&gt;Create a folder &lt;code&gt;C:\nuget&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;In PowerShell as Administrator, point NuGet at it:
&lt;/li&gt;
&lt;/ol&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight powershell"&gt;&lt;code&gt;&lt;span class="w"&gt;   &lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;Environment&lt;/span&gt;&lt;span class="p"&gt;]::&lt;/span&gt;&lt;span class="n"&gt;SetEnvironmentVariable&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"NUGET_PACKAGES"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"C:\nuget"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"Machine"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h3&gt;
  
  
  Twilio Video
&lt;/h3&gt;

&lt;p&gt;Twilio doesn't provide an official .NET MAUI library, but there are options:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Rebuild the video screens "the MAUI way"&lt;/strong&gt; with &lt;a href="https://github.com/the49ltd/The49.Twilio.Video" rel="noopener noreferrer"&gt;The49.Twilio.Video&lt;/a&gt;, a community package.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Reuse your Xamarin custom renderers&lt;/strong&gt; with just the binding projects from that repository (&lt;code&gt;The49.Twilio.Video.iOS&lt;/code&gt; and &lt;code&gt;The49.Twilio.Video.Android&lt;/code&gt;). You'll change some references and possibly some code to match Twilio Video API changes. When upgrading from an older Twilio version, replace &lt;code&gt;AddRenderer()&lt;/code&gt; with &lt;code&gt;AddSink()&lt;/code&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you reference the binding projects directly, Visual Studio may not recognize their namespaces. Reference the built DLL from the binding project instead.&lt;/p&gt;

&lt;h2&gt;
  
  
  Conclusion
&lt;/h2&gt;

&lt;p&gt;.NET MAUI is the future of cross-platform app development with .NET. The tools help with the first pass, but plan for testing and manual fixes, especially for layouts, custom renderers and third-party packages. The app stores have now made the decision for you: without a migration, a Xamarin app can't be updated. Once you're on .NET MAUI, keep it current with a yearly upgrade.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Is Xamarin still supported?
&lt;/h3&gt;

&lt;p&gt;No. Microsoft ended support for all Xamarin SDKs, including Xamarin.Forms, on May 1, 2024. There are no more bug fixes, security updates or support for new Android and iOS versions.&lt;/p&gt;

&lt;h3&gt;
  
  
  Can I still publish updates to a Xamarin app?
&lt;/h3&gt;

&lt;p&gt;Not to the app stores. Xamarin's last versions target Android API 34 and Xcode 15. Google Play now requires updates to target Android 16 (API 36), and the App Store requires builds made with Xcode 26.&lt;/p&gt;

&lt;h3&gt;
  
  
  Which .NET version should I migrate to?
&lt;/h3&gt;

&lt;p&gt;.NET 10, with .NET MAUI 10. .NET 8 and .NET 9 reach end of support on November 10, 2026, and MAUI 8 and 9 are already out of support.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is there a tool that migrates a Xamarin app automatically?
&lt;/h3&gt;

&lt;p&gt;Not completely. The GitHub Copilot modernization agent in Visual Studio, which replaced the deprecated .NET Upgrade Assistant, handles much of the project and code conversion. Layouts, custom renderers and third-party packages still need manual work and testing.&lt;/p&gt;

&lt;h3&gt;
  
  
  Do I need to rewrite my app to move to .NET MAUI?
&lt;/h3&gt;

&lt;p&gt;Usually not. .NET MAUI is based on Xamarin.Forms, so most of your C# code and much of your XAML carries over. The work is in the project structure, layouts, renderers and packages.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;SharpQuest builds and migrates cross-platform mobile apps with .NET MAUI. More on that in our &lt;a href="https://www.sharpquest.com/mobile-app-development-company/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-xamarin-maui" rel="noopener noreferrer"&gt;mobile app development&lt;/a&gt; work.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>dotnet</category>
      <category>csharp</category>
      <category>mobile</category>
      <category>maui</category>
    </item>
    <item>
      <title>Angular Best Practices for 2026: What Still Holds and What's New</title>
      <dc:creator>Piyush Bhatt</dc:creator>
      <pubDate>Thu, 29 Oct 2026 16:53:36 +0000</pubDate>
      <link>https://dev.to/piyushpbhatt/angular-best-practices-for-2026-what-still-holds-and-whats-new-2l7a</link>
      <guid>https://dev.to/piyushpbhatt/angular-best-practices-for-2026-what-still-holds-and-whats-new-2l7a</guid>
      <description>&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://www.sharpquest.com/blog/posts/2026/october/angular-best-practices-2026/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-angular-2026" rel="noopener noreferrer"&gt;SharpQuest blog&lt;/a&gt;, written by Greg Kabacinski with our Angular team.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key takeaways&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Build new apps with standalone components; migrate NgModule apps step by step with the CLI's migrations.&lt;/li&gt;
&lt;li&gt;Use signals for component state (&lt;code&gt;signal&lt;/code&gt;, &lt;code&gt;computed&lt;/code&gt;, &lt;code&gt;input()&lt;/code&gt;, &lt;code&gt;output()&lt;/code&gt;), and RxJS for events and async streams.&lt;/li&gt;
&lt;li&gt;Use the built-in control flow (&lt;code&gt;@if&lt;/code&gt;, &lt;code&gt;@for&lt;/code&gt; with &lt;code&gt;track&lt;/code&gt;) and &lt;code&gt;@defer&lt;/code&gt; to load heavy parts of a page only when needed.&lt;/li&gt;
&lt;li&gt;Make components &lt;code&gt;OnPush&lt;/code&gt;; with signals, new apps can run without zone.js.&lt;/li&gt;
&lt;li&gt;Keep the 2022 basics: small single-purpose files, Prettier, lazy-loaded routes, presentational components, and no leaked subscriptions.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;In 2022 we wrote down the &lt;a href="https://www.sharpquest.com/blog/posts/2022/september/angular-best-practices-and-core-concepts-2022/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-angular-2026" rel="noopener noreferrer"&gt;Angular best practices our team had learned&lt;/a&gt; from four years of enterprise apps. Since then Angular has changed more than in any period since Angular 2: standalone components, signals, a new template syntax and a faster build. This post covers what we do on new Angular apps in 2026, and which of our 2022 practices still hold.&lt;/p&gt;

&lt;p&gt;Our Angular work includes a &lt;a href="https://www.sharpquest.com/portfolio/graham-healthcare-group-clinician-pwa-for-homecare/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-angular-2026" rel="noopener noreferrer"&gt;progressive web app for home-care clinicians&lt;/a&gt; at a healthcare provider, and products for a technology startup and an employee-benefits software company.&lt;/p&gt;

&lt;h2&gt;
  
  
  What still holds from 2022
&lt;/h2&gt;

&lt;p&gt;These haven't changed, and we still apply them on every project:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;One thing per file.&lt;/strong&gt; One component, service or directive per file, named after what it is.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Let tools do the formatting.&lt;/strong&gt; &lt;a href="https://prettier.io/" rel="noopener noreferrer"&gt;Prettier&lt;/a&gt; with a pre-commit hook, plus ESLint through &lt;a href="https://github.com/angular-eslint/angular-eslint" rel="noopener noreferrer"&gt;angular-eslint&lt;/a&gt; (&lt;code&gt;ng add @angular-eslint/schematics&lt;/code&gt;).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Lazy-load features.&lt;/strong&gt; The first download should contain only what the first screen needs.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Smart and presentational components.&lt;/strong&gt; Screens load data and handle logic; the components inside them just take inputs and raise outputs.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Don't leak subscriptions.&lt;/strong&gt; Use the async pipe, &lt;code&gt;take(1)&lt;/code&gt;, or &lt;code&gt;takeUntilDestroyed()&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;No expensive work in templates.&lt;/strong&gt; Templates should read values, not compute them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Path aliases&lt;/strong&gt; in &lt;a href="https://www.typescriptlang.org/tsconfig/#paths" rel="noopener noreferrer"&gt;&lt;code&gt;tsconfig.json&lt;/code&gt;&lt;/a&gt; (&lt;code&gt;@core/*&lt;/code&gt;, &lt;code&gt;@shared/*&lt;/code&gt;) instead of long &lt;code&gt;../../../&lt;/code&gt; imports.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;What changed is how you do most of these. The rest of this post is the 2026 version.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Standalone components, not NgModules
&lt;/h2&gt;

&lt;p&gt;Since Angular 19, components, directives and pipes are standalone by default. A component imports what it uses directly, and the app starts with &lt;code&gt;bootstrapApplication()&lt;/code&gt; instead of an &lt;code&gt;AppModule&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;&lt;span class="c1"&gt;// main.ts&lt;/span&gt;
&lt;span class="nf"&gt;bootstrapApplication&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;App&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="na"&gt;providers&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;
    &lt;span class="nf"&gt;provideRouter&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;routes&lt;/span&gt;&lt;span class="p"&gt;),&lt;/span&gt;
    &lt;span class="nf"&gt;provideHttpClient&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nf"&gt;withInterceptors&lt;/span&gt;&lt;span class="p"&gt;([&lt;/span&gt;&lt;span class="nx"&gt;authInterceptor&lt;/span&gt;&lt;span class="p"&gt;])),&lt;/span&gt;
  &lt;span class="p"&gt;],&lt;/span&gt;
&lt;span class="p"&gt;});&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;





&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;&lt;span class="p"&gt;@&lt;/span&gt;&lt;span class="nd"&gt;Component&lt;/span&gt;&lt;span class="p"&gt;({&lt;/span&gt;
  &lt;span class="na"&gt;selector&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="dl"&gt;'&lt;/span&gt;&lt;span class="s1"&gt;app-order-list&lt;/span&gt;&lt;span class="dl"&gt;'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="na"&gt;imports&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="nx"&gt;CurrencyPipe&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nx"&gt;OrderRow&lt;/span&gt;&lt;span class="p"&gt;],&lt;/span&gt;
  &lt;span class="na"&gt;templateUrl&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="dl"&gt;'&lt;/span&gt;&lt;span class="s1"&gt;./order-list.html&lt;/span&gt;&lt;span class="dl"&gt;'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="p"&gt;})&lt;/span&gt;
&lt;span class="k"&gt;export&lt;/span&gt; &lt;span class="kd"&gt;class&lt;/span&gt; &lt;span class="nc"&gt;OrderList&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;strong&gt;Existing apps:&lt;/strong&gt; you don't need to rewrite anything. NgModules are still supported, and the two styles work side by side. &lt;a href="https://angular.dev/reference/migrations/standalone" rel="noopener noreferrer"&gt;&lt;code&gt;ng generate @angular/core:standalone&lt;/code&gt;&lt;/a&gt; migrates an app in three passes (convert components, remove unneeded modules, switch to &lt;code&gt;bootstrapApplication&lt;/code&gt;). We run it feature by feature, with a build and a test run between passes.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. The new style guide: shorter names
&lt;/h2&gt;

&lt;p&gt;Angular rewrote its &lt;a href="https://angular.dev/style-guide" rel="noopener noreferrer"&gt;style guide&lt;/a&gt; in 2025 (Angular 20). The biggest visible change: no type suffixes for components, directives and services. The CLI now generates &lt;code&gt;order-list.ts&lt;/code&gt; with class &lt;code&gt;OrderList&lt;/code&gt;, not &lt;code&gt;order-list.component.ts&lt;/code&gt; with &lt;code&gt;OrderListComponent&lt;/code&gt;. Other points:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Use the &lt;a href="https://angular.dev/guide/di" rel="noopener noreferrer"&gt;&lt;code&gt;inject()&lt;/code&gt; function&lt;/a&gt; instead of constructor injection.&lt;/li&gt;
&lt;li&gt;Keep components focused on presentation; move logic into services or plain functions.&lt;/li&gt;
&lt;li&gt;Use &lt;code&gt;protected&lt;/code&gt; for members that only the template uses.&lt;/li&gt;
&lt;li&gt;Group files by feature, not by type.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If an existing project uses suffixes, keep them there. Consistency within a codebase matters more than following the newest guide.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Signals for state
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://angular.dev/guide/signals" rel="noopener noreferrer"&gt;Signals&lt;/a&gt; are Angular's built-in reactive values. A signal holds a value, a &lt;code&gt;computed()&lt;/code&gt; signal derives one, and Angular updates exactly the parts of the page that read them:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;&lt;span class="k"&gt;export&lt;/span&gt; &lt;span class="kd"&gt;class&lt;/span&gt; &lt;span class="nc"&gt;Cart&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="nx"&gt;items&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nx"&gt;signal&lt;/span&gt;&lt;span class="o"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nx"&gt;CartItem&lt;/span&gt;&lt;span class="p"&gt;[]&lt;/span&gt;&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;&lt;span class="p"&gt;([]);&lt;/span&gt;
  &lt;span class="nx"&gt;total&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;computed&lt;/span&gt;&lt;span class="p"&gt;(()&lt;/span&gt; &lt;span class="o"&gt;=&amp;gt;&lt;/span&gt; &lt;span class="k"&gt;this&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;items&lt;/span&gt;&lt;span class="p"&gt;().&lt;/span&gt;&lt;span class="nf"&gt;reduce&lt;/span&gt;&lt;span class="p"&gt;((&lt;/span&gt;&lt;span class="nx"&gt;sum&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nx"&gt;i&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;=&amp;gt;&lt;/span&gt; &lt;span class="nx"&gt;sum&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="nx"&gt;i&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;price&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="nx"&gt;i&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;qty&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;));&lt;/span&gt;

  &lt;span class="nf"&gt;add&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;item&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nx"&gt;CartItem&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="k"&gt;this&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;items&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;update&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;items&lt;/span&gt; &lt;span class="o"&gt;=&amp;gt;&lt;/span&gt; &lt;span class="p"&gt;[...&lt;/span&gt;&lt;span class="nx"&gt;items&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nx"&gt;item&lt;/span&gt;&lt;span class="p"&gt;]);&lt;/span&gt;
  &lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;





&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight html"&gt;&lt;code&gt;&lt;span class="nt"&gt;&amp;lt;p&amp;gt;&lt;/span&gt;Total: {{ total() | currency }}&lt;span class="nt"&gt;&amp;lt;/p&amp;gt;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Calling &lt;code&gt;total()&lt;/code&gt; in the template is fine. That's how signals are read, and Angular only re-evaluates it when &lt;code&gt;items&lt;/code&gt; changes. (Our 2022 advice against functions in templates still applies to ordinary methods.)&lt;/p&gt;

&lt;p&gt;Components use signal-based &lt;a href="https://angular.dev/guide/components/inputs" rel="noopener noreferrer"&gt;inputs&lt;/a&gt;, &lt;a href="https://angular.dev/guide/components/outputs" rel="noopener noreferrer"&gt;outputs&lt;/a&gt; and queries:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;&lt;span class="k"&gt;export&lt;/span&gt; &lt;span class="kd"&gt;class&lt;/span&gt; &lt;span class="nc"&gt;LessonsList&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="nx"&gt;lessons&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nx"&gt;input&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;required&lt;/span&gt;&lt;span class="o"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nx"&gt;Lesson&lt;/span&gt;&lt;span class="p"&gt;[]&lt;/span&gt;&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;&lt;span class="p"&gt;();&lt;/span&gt;
  &lt;span class="nx"&gt;selected&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nx"&gt;output&lt;/span&gt;&lt;span class="o"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nx"&gt;Lesson&lt;/span&gt;&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;&lt;span class="p"&gt;();&lt;/span&gt;
  &lt;span class="nx"&gt;search&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;model&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="dl"&gt;''&lt;/span&gt;&lt;span class="p"&gt;);&lt;/span&gt;           &lt;span class="c1"&gt;// two-way binding&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Practices we follow:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Use &lt;code&gt;computed()&lt;/code&gt; for derived values, not &lt;code&gt;effect()&lt;/code&gt; that sets another signal.&lt;/li&gt;
&lt;li&gt;Keep &lt;code&gt;effect()&lt;/code&gt; for side effects that leave Angular: logging, local storage, a third-party chart.&lt;/li&gt;
&lt;li&gt;Update signals immutably (&lt;code&gt;update(items =&amp;gt; [...items, item])&lt;/code&gt;), so &lt;code&gt;computed()&lt;/code&gt; and &lt;code&gt;track&lt;/code&gt; see the change.&lt;/li&gt;
&lt;li&gt;Use &lt;a href="https://angular.dev/guide/signals/linked-signal" rel="noopener noreferrer"&gt;&lt;code&gt;linkedSignal()&lt;/code&gt;&lt;/a&gt; for a value that resets when another changes (for example, the selected item when the list changes).&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  4. Signals and RxJS: use both
&lt;/h2&gt;

&lt;p&gt;Signals don't replace &lt;a href="https://rxjs.dev/" rel="noopener noreferrer"&gt;RxJS&lt;/a&gt;. Our rule of thumb:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Use signals for&lt;/th&gt;
&lt;th&gt;Use RxJS for&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Component and screen state&lt;/td&gt;
&lt;td&gt;HTTP calls, websockets, server events&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Values the template reads&lt;/td&gt;
&lt;td&gt;User input over time (search-as-you-type with &lt;code&gt;debounceTime&lt;/code&gt; + &lt;code&gt;switchMap&lt;/code&gt;)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Derived values (&lt;code&gt;computed&lt;/code&gt;)&lt;/td&gt;
&lt;td&gt;Combining, retrying or cancelling async work&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Convert at the boundary with &lt;a href="https://angular.dev/ecosystem/rxjs-interop" rel="noopener noreferrer"&gt;RxJS interop&lt;/a&gt;: &lt;code&gt;toSignal(this.http.get(...))&lt;/code&gt; to show the result in a template, &lt;code&gt;toObservable(signal)&lt;/code&gt; to feed a signal into an RxJS pipeline. For loading data, &lt;a href="https://angular.dev/guide/signals/resource" rel="noopener noreferrer"&gt;&lt;code&gt;resource()&lt;/code&gt; and &lt;code&gt;httpResource()&lt;/code&gt;&lt;/a&gt; put async results straight into signals, with loading and error states built in; they're stable since Angular 22.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Built-in control flow
&lt;/h2&gt;

&lt;p&gt;Angular 17 added &lt;a href="https://angular.dev/guide/templates/control-flow" rel="noopener noreferrer"&gt;control flow blocks&lt;/a&gt; to templates, and they're now the standard. They replace &lt;code&gt;*ngIf&lt;/code&gt;, &lt;code&gt;*ngFor&lt;/code&gt; and &lt;code&gt;*ngSwitch&lt;/code&gt; (deprecated since Angular 20):&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight html"&gt;&lt;code&gt;@if (user(); as user) {
  &lt;span class="nt"&gt;&amp;lt;p&amp;gt;&lt;/span&gt;Welcome {{ user.name }}&lt;span class="nt"&gt;&amp;lt;/p&amp;gt;&lt;/span&gt;
} @else {
  &lt;span class="nt"&gt;&amp;lt;a&lt;/span&gt; &lt;span class="na"&gt;routerLink=&lt;/span&gt;&lt;span class="s"&gt;"/login"&lt;/span&gt;&lt;span class="nt"&gt;&amp;gt;&lt;/span&gt;Sign in&lt;span class="nt"&gt;&amp;lt;/a&amp;gt;&lt;/span&gt;
}

@for (order of orders(); track order.id) {
  &lt;span class="nt"&gt;&amp;lt;app-order-row&lt;/span&gt; &lt;span class="na"&gt;[order]=&lt;/span&gt;&lt;span class="s"&gt;"order"&lt;/span&gt; &lt;span class="nt"&gt;/&amp;gt;&lt;/span&gt;
} @empty {
  &lt;span class="nt"&gt;&amp;lt;p&amp;gt;&lt;/span&gt;No orders yet.&lt;span class="nt"&gt;&amp;lt;/p&amp;gt;&lt;/span&gt;
}
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;@for&lt;/code&gt; requires &lt;code&gt;track&lt;/code&gt;, which fixes the most common list performance problem from our 2022 post. Track by a unique id, not &lt;code&gt;$index&lt;/code&gt;. &lt;a href="https://angular.dev/reference/migrations/control-flow" rel="noopener noreferrer"&gt;&lt;code&gt;ng generate @angular/core:control-flow&lt;/code&gt;&lt;/a&gt; converts existing templates.&lt;/p&gt;

&lt;h2&gt;
  
  
  6. &lt;a class="mentioned-user" href="https://dev.to/defer"&gt;@defer&lt;/a&gt;: lazy-load inside a page
&lt;/h2&gt;

&lt;p&gt;Route-level lazy loading is still the first step (&lt;a href="https://angular.dev/guide/routing/loading-strategies" rel="noopener noreferrer"&gt;&lt;code&gt;loadComponent&lt;/code&gt; for one component, &lt;code&gt;loadChildren&lt;/code&gt; for a feature's routes&lt;/a&gt;). &lt;a href="https://angular.dev/guide/templates/defer" rel="noopener noreferrer"&gt;&lt;code&gt;@defer&lt;/code&gt;&lt;/a&gt; goes further and lazy-loads part of a template:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight html"&gt;&lt;code&gt;@defer (on viewport) {
  &lt;span class="nt"&gt;&amp;lt;app-sales-chart&lt;/span&gt; &lt;span class="na"&gt;[data]=&lt;/span&gt;&lt;span class="s"&gt;"sales()"&lt;/span&gt; &lt;span class="nt"&gt;/&amp;gt;&lt;/span&gt;
} @placeholder {
  &lt;span class="nt"&gt;&amp;lt;div&lt;/span&gt; &lt;span class="na"&gt;class=&lt;/span&gt;&lt;span class="s"&gt;"chart-placeholder"&lt;/span&gt;&lt;span class="nt"&gt;&amp;gt;&amp;lt;/div&amp;gt;&lt;/span&gt;
} @loading (minimum 300ms) {
  &lt;span class="nt"&gt;&amp;lt;app-spinner&lt;/span&gt; &lt;span class="nt"&gt;/&amp;gt;&lt;/span&gt;
}
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The chart's code isn't downloaded until it scrolls into view. Good candidates: charts, rich text editors, maps, and anything below the fold or behind a tab.&lt;/p&gt;

&lt;h2&gt;
  
  
  7. Change detection: OnPush and zoneless
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Make every component &lt;a href="https://angular.dev/best-practices/skipping-subtrees" rel="noopener noreferrer"&gt;&lt;code&gt;OnPush&lt;/code&gt;&lt;/a&gt;.&lt;/strong&gt; With signals and inputs, it's almost free, and it stops Angular from checking components whose data hasn't changed. Set it as the CLI default in &lt;code&gt;angular.json&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;a href="https://angular.dev/guide/zoneless" rel="noopener noreferrer"&gt;Zoneless&lt;/a&gt;.&lt;/strong&gt; Angular no longer needs zone.js to know when to update the page; signals, events and the async pipe tell it directly. Zoneless is the default for new apps since Angular 21. For an existing app, remove zone.js only after its components are OnPush and signal-based, then test the screens that update after timers or third-party callbacks.&lt;/li&gt;
&lt;li&gt;Large lists: use &lt;code&gt;@for&lt;/code&gt; with &lt;code&gt;track&lt;/code&gt;, and the &lt;a href="https://material.angular.dev/cdk/scrolling/overview" rel="noopener noreferrer"&gt;CDK's virtual scrolling&lt;/a&gt; (&lt;code&gt;cdk-virtual-scroll-viewport&lt;/code&gt;) for thousands of rows.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  8. State management: start simple
&lt;/h2&gt;

&lt;p&gt;Most business apps don't need a state library on day one. What we use:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;App size&lt;/th&gt;
&lt;th&gt;Approach&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;A few screens with their own data&lt;/td&gt;
&lt;td&gt;Signals in the component&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;State shared between screens&lt;/td&gt;
&lt;td&gt;A service holding signals (private &lt;code&gt;signal&lt;/code&gt;, public &lt;code&gt;asReadonly()&lt;/code&gt; and &lt;code&gt;computed&lt;/code&gt;)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Large apps, many teams, complex shared state&lt;/td&gt;
&lt;td&gt;
&lt;a href="https://ngrx.io/guide/signals/signal-store" rel="noopener noreferrer"&gt;NgRx Signal Store&lt;/a&gt;; classic &lt;a href="https://ngrx.io/" rel="noopener noreferrer"&gt;NgRx&lt;/a&gt; Store where an app already uses it&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;NgRx's older &lt;a href="https://ngrx.io/guide/component-store" rel="noopener noreferrer"&gt;Component Store&lt;/a&gt; is still supported, but for new code we use Signal Store.&lt;/p&gt;

&lt;h2&gt;
  
  
  9. Forms
&lt;/h2&gt;

&lt;p&gt;In existing code, use &lt;a href="https://angular.dev/guide/forms/typed-forms" rel="noopener noreferrer"&gt;typed reactive forms&lt;/a&gt; (since Angular 14): &lt;code&gt;FormControl&lt;/code&gt; catches mistakes at compile time that untyped forms only show at runtime. Use &lt;code&gt;nonNullable&lt;/code&gt; controls so &lt;code&gt;reset()&lt;/code&gt; doesn't produce &lt;code&gt;null&lt;/code&gt;. For new forms, &lt;a href="https://angular.dev/guide/forms/signals/overview" rel="noopener noreferrer"&gt;Signal Forms&lt;/a&gt; build the form model on signals and are stable since Angular 22.&lt;/p&gt;

&lt;h2&gt;
  
  
  10. Build, SSR and performance
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Build:&lt;/strong&gt; the &lt;code&gt;application&lt;/code&gt; builder (&lt;a href="https://esbuild.github.io/" rel="noopener noreferrer"&gt;esbuild&lt;/a&gt; and &lt;a href="https://vite.dev/" rel="noopener noreferrer"&gt;Vite&lt;/a&gt;) is the default. &lt;a href="https://angular.dev/tools/cli/build-system-migration" rel="noopener noreferrer"&gt;Moving an older app off the webpack builder&lt;/a&gt; usually cuts build times a lot. &lt;code&gt;ng update&lt;/code&gt; offers the migration.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Server-side rendering:&lt;/strong&gt; for public, search-indexed pages, &lt;a href="https://angular.dev/guide/ssr" rel="noopener noreferrer"&gt;&lt;code&gt;@angular/ssr&lt;/code&gt;&lt;/a&gt; with hydration (and incremental hydration with &lt;code&gt;@defer&lt;/code&gt;) gives a faster first paint and better SEO. Internal business apps behind a login usually don't need it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Images:&lt;/strong&gt; &lt;a href="https://angular.dev/guide/image-optimization" rel="noopener noreferrer"&gt;&lt;code&gt;NgOptimizedImage&lt;/code&gt;&lt;/a&gt; (&lt;code&gt;ngSrc&lt;/code&gt;) sets sizes, lazy-loads below the fold and prioritizes the main image.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Budgets:&lt;/strong&gt; keep the bundle budgets in &lt;code&gt;angular.json&lt;/code&gt; and fail the build when they're exceeded.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  11. Testing
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Unit tests:&lt;/strong&gt; &lt;a href="https://vitest.dev/" rel="noopener noreferrer"&gt;Vitest&lt;/a&gt; is the default test runner for new CLI projects (see Angular's &lt;a href="https://angular.dev/guide/testing" rel="noopener noreferrer"&gt;testing guide&lt;/a&gt;); Karma is still supported for existing ones. Test components through their template (what the user sees), not their private methods.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Signals in tests:&lt;/strong&gt; set inputs with &lt;code&gt;fixture.componentRef.setInput()&lt;/code&gt;, then check the rendered output.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;End-to-end:&lt;/strong&gt; &lt;a href="https://playwright.dev/" rel="noopener noreferrer"&gt;Playwright&lt;/a&gt; or &lt;a href="https://www.cypress.io/" rel="noopener noreferrer"&gt;Cypress&lt;/a&gt; for the main user flows. Protractor is gone.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  12. Security basics
&lt;/h2&gt;

&lt;p&gt;Angular &lt;a href="https://angular.dev/best-practices/security" rel="noopener noreferrer"&gt;escapes values in templates by default&lt;/a&gt;, so most cross-site scripting needs a bypass to happen. Don't use &lt;code&gt;bypassSecurityTrust*&lt;/code&gt; with user data, keep HTTP calls on &lt;code&gt;HttpClient&lt;/code&gt; so its XSRF support works, and set a Content Security Policy.&lt;/p&gt;

&lt;h2&gt;
  
  
  13. Keep up with releases
&lt;/h2&gt;

&lt;p&gt;Since Angular 22, a new major version comes out every 12 months (it was every 6 months before), and each one gets about 24 months of support: 12 months of active support, then 12 months of long-term support (see &lt;a href="https://angular.dev/reference/releases" rel="noopener noreferrer"&gt;versioning and releases&lt;/a&gt;). Upgrade at least once a year; skipping several versions is how a routine &lt;code&gt;ng update&lt;/code&gt; becomes a project. &lt;a href="https://angular.dev/cli/update" rel="noopener noreferrer"&gt;&lt;code&gt;ng update&lt;/code&gt;&lt;/a&gt; runs migrations for you, one major version at a time, and Angular's &lt;a href="https://angular.dev/update-guide" rel="noopener noreferrer"&gt;Update guide&lt;/a&gt; lists the manual steps for each version.&lt;/p&gt;

&lt;h2&gt;
  
  
  Frequently asked questions
&lt;/h2&gt;

&lt;h3&gt;
  
  
  What are the most important Angular best practices in 2026?
&lt;/h3&gt;

&lt;p&gt;Standalone components, signals for state, the built-in control flow with &lt;code&gt;track&lt;/code&gt;, &lt;code&gt;@defer&lt;/code&gt; and route lazy loading, OnPush change detection, and typed or signal-based forms. The basics still apply: small files, Prettier and ESLint, and presentational components.&lt;/p&gt;

&lt;h3&gt;
  
  
  Should I still use NgModules?
&lt;/h3&gt;

&lt;p&gt;Not for new code. Existing NgModule apps are supported and can move to standalone gradually with &lt;code&gt;ng generate @angular/core:standalone&lt;/code&gt;.&lt;/p&gt;

&lt;h3&gt;
  
  
  Do signals replace RxJS?
&lt;/h3&gt;

&lt;p&gt;No. Use signals for state the template reads and RxJS for async streams such as HTTP calls and user input. &lt;code&gt;toSignal()&lt;/code&gt; and &lt;code&gt;toObservable()&lt;/code&gt; connect them.&lt;/p&gt;

&lt;h3&gt;
  
  
  Do I need NgRx?
&lt;/h3&gt;

&lt;p&gt;Usually not at the start. Signals in components and services cover most business apps. Consider NgRx Signal Store when shared state grows across many screens or teams.&lt;/p&gt;

&lt;h3&gt;
  
  
  How often should we upgrade Angular?
&lt;/h3&gt;

&lt;p&gt;At least once a year. A new major version comes out every 12 months and is supported for about 24 months, and &lt;code&gt;ng update&lt;/code&gt; upgrades one major version at a time.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is it worth migrating an existing Angular app to signals and standalone?
&lt;/h3&gt;

&lt;p&gt;Yes, gradually. The CLI migrations do most of the mechanical work; plan the time for testing. Start with the screens you're already changing.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;SharpQuest builds Angular front ends for business applications and upgrades older Angular apps, including AngularJS, to current versions. More on that in our &lt;a href="https://www.sharpquest.com/application-modernization/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=blog-angular-2026" rel="noopener noreferrer"&gt;application modernization&lt;/a&gt; work.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>angular</category>
      <category>typescript</category>
      <category>webdev</category>
      <category>javascript</category>
    </item>
  </channel>
</rss>
