{"id":15792,"date":"2026-11-17T17:36:41","date_gmt":"2026-11-17T17:36:41","guid":{"rendered":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/"},"modified":"2026-11-17T17:36:42","modified_gmt":"2026-11-17T17:36:42","slug":"web-portal-to-mobile-app-in-24-hours-ai","status":"publish","type":"post","link":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/","title":{"rendered":"From Web Portal to Mobile App in 24 Hours: What Changes When AI Learns Your Codebase"},"content":{"rendered":"<p>Have you ever wondered what happens when a client with a mature web portal already running in production asks: <em>&#8220;what about the app?&#8221;<\/em> The traditional answer usually comes with a long schedule: assemble a mobile team, redesign flows, reimplement business rules that took years to stabilize, and hope the two platforms don&#8217;t start diverging in the very first sprint.<\/p>\n<p>We got exactly that question. And our answer was different.<\/p>\n<p>Picture the scene: on Monday the app is started, and twenty-four hours later, in five commits, it is navigable on iOS and Android: 17 screens, 11 of the 12 features it has today, 112 passing tests and the native projects configured for all three environments (dev, staging and production).<\/p>\n<p>It sounds like magic, but it isn&#8217;t. It also wasn&#8217;t a genius prompt along the lines of <em>&#8220;Claude, convert this portal to React Native&#8221;<\/em>. It was the opposite: we spent more time teaching the AI to work in our repository (the folder where the project&#8217;s code lives) than asking it for code. In this post I&#8217;ll share what we built and what it changes for those who buy and those who deliver software.<\/p>\n<h2>The starting point: a portal that already worked<\/h2>\n<p>The client is a global political risk analysis company. Its web portal delivers articles, events, conference calls, daily emails and dashboards to subscribers, and was already integrated with a headless CMS, CRM, email platform, analytics and error monitoring. It comes to about 179,000 lines across front end and back end, 32 pages and 159 endpoints.<\/p>\n<p>That completely changed the question. It stopped being <em>&#8220;how do we build an app?&#8221;<\/em> and became &#8220;how do we translate a portal that already works?&#8221;.<\/p>\n<p>The business rules, the data format, the API addresses and even the design&#8217;s colors and fonts were already settled. The job was to carry all of that over to another platform without losing anything along the way. And that is exactly where the naive approach to AI breaks down.<\/p>\n<h2>Why &#8220;just convert it for me&#8221; doesn&#8217;t work<\/h2>\n<p>Asking an AI to convert an entire project fails fast. Worse: it fails silently. We saw four failure modes, and they all share the same cause:<\/p>\n<ul>\n<li><strong>Invented data:<\/strong> the AI creates a field the API doesn&#8217;t return, and the bug only shows up with the app running.<\/li>\n<li><strong>Wrong addresses:<\/strong> the AI copies the address the website uses, but the real server&#8217;s address is a different one. Anyone who only looks at the website gets it wrong.<\/li>\n<li><strong>Architecture drift:<\/strong> every screen is born with a different pattern, because no written pattern exists.<\/li>\n<li><strong>Visual drift:<\/strong> a blue here, an almost identical blue there, and suddenly the app no longer looks like the product.<\/li>\n<\/ul>\n<p>The common cause? The AI fills gaps with something plausible. When it doesn&#8217;t know where the information is, it doesn&#8217;t stop: it invents. And it invents well enough to go unnoticed.<\/p>\n<p>So the right question wasn&#8217;t how to make the AI write faster. It was how to remove its need to guess.<\/p>\n<h2>The repository that teaches the AI<\/h2>\n<p>Think of an AI assistant as a brilliant, extremely fast senior developer who joined the team today&#8230; and wakes up with amnesia every morning. Every session starts from zero. They don&#8217;t remember yesterday&#8217;s conversation, last week&#8217;s architecture decision or why that screen is built the way it is.<\/p>\n<p>With a colleague like that, you don&#8217;t repeat everything in every conversation. You leave notes. You leave the manual on the desk. And for the things that really can&#8217;t go wrong, you install a lock on the door.<\/p>\n<p>That&#8217;s what we did with Claude Code. All told, the repository has about 14,400 lines of instructions and guardrails for 58,000 lines of code, roughly one line of spec for every four lines of code. This setup is organized into six layers:<\/p>\n<ul>\n<li><strong>Constitution:<\/strong> two CLAUDE.md files, one at the root of the repository and another inside the mobile app, with the non-negotiable rules: layers, direction of data flow, naming, testing strategy. They are loaded in every session without anyone asking.<\/li>\n<li><strong>Spec:<\/strong> 20 documents covering architecture, error handling, testing and the design system, plus feature plans and tracking tables. One detail that made a difference: the testing doc specifies what <em>not<\/em> to test. Without that limit, the AI generates tests for everything, and the suite becomes a cost instead of a signal.<\/li>\n<li><strong>Procedures:<\/strong> 17 <em>skills<\/em>, reusable recipes that Claude loads on its own when the task matches the description. The main one, port-web-feature, teaches how to translate a feature from the web to the app.<\/li>\n<li><strong>Reviewers:<\/strong> 8 <em>subagents<\/em>, separate instances with their own context. The mobile reviewer checks the diff against the documented checklist and is explicitly instructed to stay quiet when there is no violation. A reviewer that always speaks up is a reviewer nobody reads.<\/li>\n<li><strong>Pipelines:<\/strong> 16 <em>slash commands<\/em> that chain skills in a mandatory order.<\/li>\n<li><strong>Guardrails:<\/strong> 7 <em>hooks<\/em>, scripts that run in the tool&#8217;s lifecycle and can block an action before it happens.<\/li>\n<\/ul>\n<p><img decoding=\"async\" class=\"alignnone size-medium wp-image-15795\" src=\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/01-six-layers-600x339.png\" alt=\"01-six-layers | | Cheesecake Labs\" width=\"600\" height=\"339\" srcset=\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/01-six-layers-600x339.png 600w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/01-six-layers-1200x677.png 1200w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/01-six-layers-768x433.png 768w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/01-six-layers-1536x867.png 1536w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/01-six-layers-760x429.png 760w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/01-six-layers.png 1616w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/><\/p>\n<p><strong>Fig. 1<\/strong> The six layers. Only the last one doesn&#8217;t depend on the model obeying: the five above are text it reads; the hook is a script with an exit code.<\/p>\n<p>The most important design decision in the project was choosing which layer each rule lives in. In short: the doc reminds, the skill teaches, the subagent isolates, the command orders and the hook blocks. An important rule that lives only in a doc can be ignored in a bad session. The same rule in a hook can&#8217;t.<\/p>\n<h2>What these layers do day to day<\/h2>\n<p>In practice, the team feels four effects:<\/p>\n<ul>\n<li><strong>The right color without pixel-by-pixel comparison.<\/strong> The app mirrors 76 design tokens from the portal, and a color outside those tokens is blocked the moment the code is saved. Nobody had to compare screens against Figma to know the color was right.<\/li>\n<li><strong>The ticket arrives on its own.<\/strong> Just mention the Jira ticket number, or not even that: the branch name is enough. Claude fetches the description and the parent ticket, cross-references them with the code and returns a plan before writing any fix.<\/li>\n<li><strong>The team&#8217;s tools within reach.<\/strong> There are seven integrations: Figma, Jira, Sentry, PostHog, Lovable, GitHub and the local development database. The remote ones use each person&#8217;s own account.<\/li>\n<li><strong>Reading is free, writing needs approval, destroying is forbidden.<\/strong> Claude runs tests, reads tickets and investigates errors without interrupting anyone. But it never pushes, installs a dependency or resolves a Sentry issue without someone approving.<\/li>\n<\/ul>\n<h2>Tests that run on their own: a robot that uses the app<\/h2>\n<p>Code written fast needs verification at the same speed. That&#8217;s why the project&#8217;s testing strategy starts with what the user does, not with the code.<\/p>\n<p>We use Maestro, a tool that opens the app on an iPhone simulator or an Android emulator and taps through the screens the way a person would: it signs in with a test user, opens an article, favorites it, saves it to read offline, searches in the menu and checks that the right screen appeared. Each script is a short text file, versioned alongside the code.<\/p>\n<p>Today there are 34 journeys across 15 areas of the app, including login, home, search, events, favorites, downloads, offline mode and the audio and video player. What makes this part of the workflow, and not a loose step at the end:<\/p>\n<ul>\n<li><strong>The AI writes the scripts.<\/strong> When a screen is ready, a skill builds the script for that journey following a convention: every important element on the screen has a predictable identifier. If the script points to something the screen doesn&#8217;t have yet, the report flags it.<\/li>\n<li><strong>The AI also runs them.<\/strong> With the app open in the simulator, Claude runs the script, reads the result and fixes what broke, instead of handing in the screen and waiting for someone to discover the problem.<\/li>\n<li><strong>The test targets the user, not the code.<\/strong> Simple screens don&#8217;t get unit tests. The script that actually uses the app covers what matters, and code tests stay only where there is business logic.<\/li>\n<li><strong>The scripts leave no trace.<\/strong> The favorites script clears the article before it starts and undoes the favorite at the end. If a run dies halfway, the next one corrects itself.<\/li>\n<li><strong>When it fails, it points to the cause.<\/strong> The script checks preconditions, such as the test account having access to the content. That way the error says &#8220;access missing&#8221; and not &#8220;the screen broke&#8221;.<\/li>\n<li><strong>The status doc doesn&#8217;t lie.<\/strong> A feature is only marked as covered by an end-to-end test once the script has passed. A script that has been written but not yet run is recorded as such.<\/li>\n<\/ul>\n<p>One use that paid off a lot: before releasing the Android build with code obfuscated by R8 (the optimization that reduces the app&#8217;s size), the release plan required running the full suite on an emulator and confirming that everything that passed without obfuscation kept passing.<\/p>\n<p>The robot has limits, and we record them. It can&#8217;t hear a device&#8217;s audio, can&#8217;t receive push notifications and can&#8217;t see the lock screen. Those points go into a manual checklist, one line per item. It&#8217;s the division of labor that makes sense: the robot covers what repeats, and a person covers what only a real device shows.<\/p>\n<h2>One command, five steps: how a web feature becomes an app feature<\/h2>\n<p>With all of this in place, porting a feature is a single command, for example \/port bookmarks, which triggers five mandatory steps, in order, with no shortcuts:<\/p>\n<ol>\n<li><strong>Pre-flight:<\/strong> a curated map answers <em>&#8220;where does the web version of this live?&#8221;<\/em> If the feature isn&#8217;t on the map, the command stops and asks instead of guessing.<\/li>\n<li><strong>Logic port:<\/strong> the skill translates the domain from the web to the app&#8217;s structure and rewrites the tests. The mapping plan goes through human approval before any file is generated.<\/li>\n<li><strong>Copy:<\/strong> the translation keys are harvested from the portal itself, not invented.<\/li>\n<li><strong>Review:<\/strong> the reviewer subagent runs against the diff. The findings are never fixed automatically, because they may be legitimate exceptions.<\/li>\n<li><strong>Tracking:<\/strong> a status table records what was actually achieved. The doc can&#8217;t be more optimistic than the code.<\/li>\n<\/ol>\n<p><img decoding=\"async\" class=\"alignnone size-medium wp-image-15797\" src=\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/02-hook-lifecycle-600x382.png\" alt=\"02-hook-lifecycle | | Cheesecake Labs\" width=\"600\" height=\"382\" srcset=\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/02-hook-lifecycle-600x382.png 600w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/02-hook-lifecycle-1200x763.png 1200w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/02-hook-lifecycle-768x489.png 768w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/02-hook-lifecycle-1536x977.png 1536w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/02-hook-lifecycle-760x483.png 760w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/02-hook-lifecycle.png 1616w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/><\/p>\n<p><strong>Fig. 2<\/strong> \/port takes the person out of where they would only type boilerplate and puts them at the three points where the decision is theirs.<\/p>\n<p>Within these steps, four things carry over from the web to the app without anyone retyping them: the data types are copied from the portal, the real backend addresses come from the portal&#8217;s code, the design tokens are mirrored and the interface text is harvested from the portal itself. None of them depends on the model&#8217;s memory.<\/p>\n<p>The gain here isn&#8217;t just speed: it&#8217;s that the web and the app don&#8217;t diverge. Same types, same endpoints, same variant names in the components, same design tokens. Two platforms, one source of truth.<\/p>\n<p>&#8220;Stop and ask&#8221; is a first-class outcome. Almost all the damage AI does to code comes from filling gaps with something plausible.<\/p>\n<h2>The numbers: 24 hours, 29 days, three months<\/h2>\n<p>Every number in this post came from the repository itself, not from estimates.<\/p>\n<p>Three calendar months passed between the first commit and the end of August. But the app was never the only thing on the table. Right after the first day, the focus went back to the web portal, and from June to the end of July the mobile app ran in the background while the portal headed for go-live: in that period there were 35 commits on the app against 190 on the portal. The app cost 29 days of work with commits, and two of them produced something navigable.<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-medium wp-image-15799\" src=\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/03-jira-flow-600x351.png\" alt=\"03-jira-flow | | Cheesecake Labs\" width=\"600\" height=\"351\" srcset=\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/03-jira-flow-600x351.png 600w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/03-jira-flow-1200x701.png 1200w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/03-jira-flow-768x449.png 768w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/03-jira-flow-1536x898.png 1536w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/03-jira-flow-760x444.png 760w, https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/03-jira-flow.png 1636w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/><\/p>\n<p><strong>Fig. 3<\/strong> Commits per day on the app, from May 25 to August 27, 2026: 112 commits over 29 days. The peak of 14 is the day the two big PRs landed in develop.<\/p>\n<p>The most interesting part is what happened after the first day:<\/p>\n<ul>\n<li>The code grew from 17,000 to 58,000 lines (3.4\u00d7), but product features went from 11 to 12. Everything else was depth: login became five screens (forgot password, reset, set, magic link), publications gained a detail view per type, and a media player arrived with background playback, lock screen controls and Picture-in-Picture.<\/li>\n<li>Pull requests shrank over time: from 665 files in the PR that brought the app in to 12 files in the last optimization PR. It&#8217;s the sign that the foundation started carrying the weight.<\/li>\n<li>For comparison, the web front end has about 78,000 lines built over years. The app reached 75% of that volume in one quarter.<\/li>\n<li>In the home stretch, a size optimization cut 24% of the download and 26% of the installed size, measured on a real build, before and after.<\/li>\n<\/ul>\n<p>The first day didn&#8217;t produce a throwaway prototype. It produced the right skeleton.<\/p>\n<p>A lesson that applies to any team working this way: with AI accelerating the writing, testing like a user carries even more weight. Types, lint and automated tests show that the code is consistent with itself; only someone navigating the screens, in the simulator or on a real device, shows that the experience works. That&#8217;s why this validation has to happen with every delivery, not just at the end.<\/p>\n<h2>Conclusion: a repository that teaches, an AI that learns<\/h2>\n<p>A 58,000-line app in three months, with one lead developer, is real and reproducible. But the number I like most in this project isn&#8217;t that one. It&#8217;s the ratio: one line of spec for every four lines of code. That layer, with 14,000 lines of conventions, guardrails and integrations, is what turned a generic model into a colleague who knows the project inside out.<\/p>\n<p>And the best part? The effect is cumulative. At the start, every feature still debated architecture, layers and patterns. By the end, pull requests had dropped from 665 to 12 files, because the decisions were already written down and nobody had to renegotiate them with every delivery. Every rule that becomes a hook, every procedure that becomes a skill and every map updated by the process itself makes the next feature cheaper than the last. Ordinary documentation ages forgotten in a wiki. This spec is read in every session and corrected whenever the code contradicts it, and that&#8217;s why it earns interest.<\/p>\n<p>Looking beyond this app, three things change in how software gets made:<\/p>\n<ul>\n<li><strong>The bottleneck moves:<\/strong> when implementing gets cheap, deciding gets relatively expensive. An ambiguous requirement costs more today than it used to, because implementation is no longer the slow part. Writing clearly what you want to build has become engineering work, not bureaucracy.<\/li>\n<li><strong>The developer&#8217;s role changes:<\/strong> less time typing boilerplate and more time where human judgment is irreplaceable: approving plans, deciding scope, judging review findings and choosing which layer each rule should live in.<\/li>\n<li><strong>The team&#8217;s knowledge becomes an asset:<\/strong> what used to live in the heads of the people who knew the project is now in versioned files that anyone, or any AI session, reads in the first minute. Onboarding stops being &#8220;ask whoever knows&#8221;.<\/li>\n<\/ul>\n<p>And the judgment is still ours. On more than one occasion, reading the code showed that copying the web would make for a worse experience on a phone, and the decision to do it differently was a person&#8217;s, recorded with a date and a reason. Feature parity is not experience parity, and no AI makes that call on its own.<\/p>\n<p>None of this is exclusive to React Native or to this client. CLAUDE.md, skills, hooks, MCPs and permissions work on any stack: what changes from one project to another is the content of the rules, not the structure. That structure is what we want to carry into our next projects, whether a new app or the modernization of a system that already exists.<\/p>\n<p>In the end, we didn&#8217;t build an app with AI. We built a way of working in which the AI understands the project almost as well as the people on it, and the app was just the first thing to come off that production line. AI, finally, adapting to the way the team works, and not the other way around.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams that buy and build software.<\/p>\n","protected":false},"author":96,"featured_media":15802,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_yoast_wpseo_title":"Web Portal to Mobile App in 24 Hours with Claude Code","_yoast_wpseo_metadesc":"How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams.","_yoast_wpseo_meta-robots-noindex":"","_yoast_wpseo_canonical":"","ai_summary":"","ai_summary_en":"","ai_summary_pt-br":"","footnotes":"","ckl_wpml_lang":"","ckl_wpml_source_id":0,"ckl_wpml_status":""},"categories":[1425,1422],"tags":[1308,1393,54,1201],"class_list":["post-15792","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-product-engineering","category-ai-implementation","tag-ai-development","tag-claude-code","tag-tag-mobile-app-development","tag-react-native"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.1.1 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Web Portal to Mobile App in 24 Hours with Claude Code<\/title>\n<meta name=\"description\" content=\"How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Web Portal to Mobile App in 24 Hours with Claude Code\" \/>\n<meta property=\"og:description\" content=\"How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/\" \/>\n<meta property=\"og:site_name\" content=\"Cheesecake Labs\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/cheesecakelabs\" \/>\n<meta property=\"article:published_time\" content=\"2026-11-17T17:36:41+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-11-17T17:36:42+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1536\" \/>\n\t<meta property=\"og:image:height\" content=\"689\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Cheesecake Labs\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@cheesecakelabs\" \/>\n<meta name=\"twitter:site\" content=\"@cheesecakelabs\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"13 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/\"},\"author\":{\"name\":\"Lucas Magnus\"},\"headline\":\"From Web Portal to Mobile App in 24 Hours: What Changes When AI Learns Your Codebase\",\"datePublished\":\"2026-11-17T17:36:41+00:00\",\"dateModified\":\"2026-11-17T17:36:42+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/\"},\"wordCount\":2654,\"publisher\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/#organization\"},\"image\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png\",\"keywords\":[\"ai development\",\"claude code\",\"mobile app development\",\"react native\"],\"articleSection\":[\"Product Engineering\",\"AI Implementation\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/\",\"url\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/\",\"name\":\"Web Portal to Mobile App in 24 Hours with Claude Code\",\"isPartOf\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png\",\"datePublished\":\"2026-11-17T17:36:41+00:00\",\"dateModified\":\"2026-11-17T17:36:42+00:00\",\"description\":\"How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams.\",\"breadcrumb\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage\",\"url\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png\",\"contentUrl\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png\",\"width\":1536,\"height\":689,\"caption\":\"Cover \u2013 Site | | Cheesecake Labs\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/cheesecakelabs.com\/blog\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"From Web Portal to Mobile App in 24 Hours: What Changes When AI Learns Your Codebase\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/#website\",\"url\":\"https:\/\/cheesecakelabs.com\/blog\/\",\"name\":\"Cheesecake Labs\",\"description\":\"AI Implementation, Data and Product Engineering\",\"publisher\":{\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/cheesecakelabs.com\/blog\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/#organization\",\"name\":\"Cheesecake Labs\",\"alternateName\":\"Cheesecake Labs Inc\",\"url\":\"https:\/\/cheesecakelabs.com\/\",\"logo\":{\"@type\":\"ImageObject\",\"@id\":\"https:\/\/cheesecakelabs.com\/#logo\",\"url\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2022\/06\/cheesecake-labs-1.png\",\"contentUrl\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2022\/06\/cheesecake-labs-1.png\",\"caption\":\"Cheesecake Labs\",\"inLanguage\":\"en\"},\"image\":{\"@type\":\"ImageObject\",\"@id\":\"https:\/\/cheesecakelabs.com\/#primary-image\",\"url\":\"https:\/\/ckl-website-v4-strapi-prod.s3.us-east-2.amazonaws.com\/ai_software_development_company_83fb512983.webp\",\"contentUrl\":\"https:\/\/ckl-website-v4-strapi-prod.s3.us-east-2.amazonaws.com\/ai_software_development_company_83fb512983.webp\",\"width\":1920,\"height\":1080,\"caption\":\"Cheesecake Labs \u2014 AI, Data & Blockchain software development services\",\"inLanguage\":\"en\"},\"sameAs\":[\"https:\/\/www.facebook.com\/cheesecakelabs\",\"https:\/\/x.com\/cheesecakelabs\",\"https:\/\/www.instagram.com\/cheesecakelabs\/\",\"https:\/\/www.linkedin.com\/company\/cheesecake-labs\/\",\"https:\/\/www.youtube.com\/channel\/UCdGEQ5AHJcmIlaOaI5fGGVA\",\"https:\/\/clutch.co\/profile\/cheesecake-labs\",\"https:\/\/www.behance.net\/cheesecakelabs\",\"https:\/\/dribbble.com\/cheesecakelabs\",\"https:\/\/www.designrush.com\/agency\/profile\/cheesecake-labs\",\"https:\/\/www.g2.com\/products\/cheesecake-labs\/reviews\"],\"description\":\"Cheesecake Labs is a software development studio that designs and builds custom digital products \u2014 web, mobile, and platforms \u2014 combining product design and high-performance engineering.\",\"foundingDate\":\"2013\"},{\"@type\":\"Person\",\"name\":\"Lucas Magnus\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/cheesecakelabs.com\/blog\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/01\/lucas-magnus-perfil.png\",\"contentUrl\":\"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/01\/lucas-magnus-perfil.png\",\"caption\":\"Lucas Magnus\"},\"url\":\"https:\/\/cheesecakelabs.com\/blog\/autor\/lucas-magnus\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Web Portal to Mobile App in 24 Hours with Claude Code","description":"How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/","og_locale":"en_US","og_type":"article","og_title":"Web Portal to Mobile App in 24 Hours with Claude Code","og_description":"How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams.","og_url":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/","og_site_name":"Cheesecake Labs","article_publisher":"https:\/\/www.facebook.com\/cheesecakelabs","article_published_time":"2026-11-17T17:36:41+00:00","article_modified_time":"2026-11-17T17:36:42+00:00","og_image":[{"width":1536,"height":689,"url":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png","type":"image\/png"}],"author":"Cheesecake Labs","twitter_card":"summary_large_image","twitter_creator":"@cheesecakelabs","twitter_site":"@cheesecakelabs","twitter_misc":{"Written by":null,"Est. reading time":"13 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#article","isPartOf":{"@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/"},"author":{"name":"Lucas Magnus"},"headline":"From Web Portal to Mobile App in 24 Hours: What Changes When AI Learns Your Codebase","datePublished":"2026-11-17T17:36:41+00:00","dateModified":"2026-11-17T17:36:42+00:00","mainEntityOfPage":{"@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/"},"wordCount":2654,"publisher":{"@id":"https:\/\/cheesecakelabs.com\/blog\/#organization"},"image":{"@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage"},"thumbnailUrl":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png","keywords":["ai development","claude code","mobile app development","react native"],"articleSection":["Product Engineering","AI Implementation"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/","url":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/","name":"Web Portal to Mobile App in 24 Hours with Claude Code","isPartOf":{"@id":"https:\/\/cheesecakelabs.com\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage"},"image":{"@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage"},"thumbnailUrl":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png","datePublished":"2026-11-17T17:36:41+00:00","dateModified":"2026-11-17T17:36:42+00:00","description":"How we took a mature web portal to a navigable iOS and Android app in 24 hours by teaching Claude Code the project first, and what it changes for teams.","breadcrumb":{"@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#primaryimage","url":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png","contentUrl":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/11\/Cover-%E2%80%93-Site.png","width":1536,"height":689,"caption":"Cover \u2013 Site | | Cheesecake Labs"},{"@type":"BreadcrumbList","@id":"https:\/\/cheesecakelabs.com\/blog\/web-portal-to-mobile-app-in-24-hours-ai\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/cheesecakelabs.com\/blog\/"},{"@type":"ListItem","position":2,"name":"From Web Portal to Mobile App in 24 Hours: What Changes When AI Learns Your Codebase"}]},{"@type":"WebSite","@id":"https:\/\/cheesecakelabs.com\/blog\/#website","url":"https:\/\/cheesecakelabs.com\/blog\/","name":"Cheesecake Labs","description":"AI Implementation, Data and Product Engineering","publisher":{"@id":"https:\/\/cheesecakelabs.com\/blog\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/cheesecakelabs.com\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/cheesecakelabs.com\/blog\/#organization","name":"Cheesecake Labs","alternateName":"Cheesecake Labs Inc","url":"https:\/\/cheesecakelabs.com\/","logo":{"@type":"ImageObject","@id":"https:\/\/cheesecakelabs.com\/#logo","url":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2022\/06\/cheesecake-labs-1.png","contentUrl":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2022\/06\/cheesecake-labs-1.png","caption":"Cheesecake Labs","inLanguage":"en"},"image":{"@type":"ImageObject","@id":"https:\/\/cheesecakelabs.com\/#primary-image","url":"https:\/\/ckl-website-v4-strapi-prod.s3.us-east-2.amazonaws.com\/ai_software_development_company_83fb512983.webp","contentUrl":"https:\/\/ckl-website-v4-strapi-prod.s3.us-east-2.amazonaws.com\/ai_software_development_company_83fb512983.webp","width":1920,"height":1080,"caption":"Cheesecake Labs \u2014 AI, Data & Blockchain software development services","inLanguage":"en"},"sameAs":["https:\/\/www.facebook.com\/cheesecakelabs","https:\/\/x.com\/cheesecakelabs","https:\/\/www.instagram.com\/cheesecakelabs\/","https:\/\/www.linkedin.com\/company\/cheesecake-labs\/","https:\/\/www.youtube.com\/channel\/UCdGEQ5AHJcmIlaOaI5fGGVA","https:\/\/clutch.co\/profile\/cheesecake-labs","https:\/\/www.behance.net\/cheesecakelabs","https:\/\/dribbble.com\/cheesecakelabs","https:\/\/www.designrush.com\/agency\/profile\/cheesecake-labs","https:\/\/www.g2.com\/products\/cheesecake-labs\/reviews"],"description":"Cheesecake Labs is a software development studio that designs and builds custom digital products \u2014 web, mobile, and platforms \u2014 combining product design and high-performance engineering.","foundingDate":"2013"},{"@type":"Person","name":"Lucas Magnus","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/cheesecakelabs.com\/blog\/#\/schema\/person\/image\/","url":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/01\/lucas-magnus-perfil.png","contentUrl":"https:\/\/ckl-website-static.s3.amazonaws.com\/wp-content\/uploads\/2026\/01\/lucas-magnus-perfil.png","caption":"Lucas Magnus"},"url":"https:\/\/cheesecakelabs.com\/blog\/autor\/lucas-magnus\/"}]}},"_links":{"self":[{"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/posts\/15792","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/users\/96"}],"replies":[{"embeddable":true,"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/comments?post=15792"}],"version-history":[{"count":2,"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/posts\/15792\/revisions"}],"predecessor-version":[{"id":15801,"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/posts\/15792\/revisions\/15801"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/media\/15802"}],"wp:attachment":[{"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/media?parent=15792"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/categories?post=15792"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/cheesecakelabs.com\/blog\/wp-json\/wp\/v2\/tags?post=15792"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}