<?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: Faizal Sukaimi</title>
    <description>The latest articles on DEV Community by Faizal Sukaimi (@faizalsukaimi).</description>
    <link>https://dev.to/faizalsukaimi</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%2F1687037%2F06e0ff80-7cc2-44ef-bbac-65557057e591.jpg</url>
      <title>DEV Community: Faizal Sukaimi</title>
      <link>https://dev.to/faizalsukaimi</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/faizalsukaimi"/>
    <language>en</language>
    <item>
      <title>9 questions to ask before choosing a shadcn/ui block library (2026)</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Mon, 16 Nov 2026 10:59:29 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/9-questions-to-ask-before-choosing-a-shadcnui-block-library-2026-4hj0</link>
      <guid>https://dev.to/faizalsukaimi/9-questions-to-ask-before-choosing-a-shadcnui-block-library-2026-4hj0</guid>
      <description>&lt;p&gt;There are a lot of shadcn/ui libraries now. Most look good in a screenshot. The differences show up later: when your coding agent invents props, when a licence blocks client work, or when you need a data grid and the library stops at buttons.&lt;/p&gt;

&lt;p&gt;These are the nine questions I'd ask before picking one. They work for any library. I'm on the ReUI team, so ReUI's own answers are in one section at the end, kept apart from the questions.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Does it extend shadcn/ui, or replace it?
&lt;/h2&gt;

&lt;p&gt;shadcn/ui works because the code lands in your repo and you own it. A library should keep that deal. Check two things. Does the CLI copy source into your project, or do you import a package you can't edit? Does it read your existing theme, or bring its own?&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Does it fit the primitives you already use?
&lt;/h2&gt;

&lt;p&gt;New shadcn/ui projects default to Base UI since July 2026. Radix UI is still supported, and React Aria arrived the same month. A block built for one may not drop cleanly into another. Check which ones the library supports before you install anything.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Can your coding agent use it without guessing?
&lt;/h2&gt;

&lt;p&gt;If you build with an agent, it will write UI from memory unless you give it something better. Memory means made-up props and a different look each time. Look for a registry the agent can install from and an MCP server that hands it the real component APIs.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Does it cover the hard components?
&lt;/h2&gt;

&lt;p&gt;Buttons and cards are easy, and shadcn/ui already has them. The time goes into data grids, filters, kanban boards, calendars and gantt charts. Check whether the library ships these, and whether they're free.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. What is free, and under which licence?
&lt;/h2&gt;

&lt;p&gt;"Free" can mean MIT, free for personal use only, or free behind a sign-up. Read the licence line, not the badge.&lt;/p&gt;

&lt;h2&gt;
  
  
  6. If you pay, what are the terms?
&lt;/h2&gt;

&lt;p&gt;Three things to check: subscription or one-time, a fee per project or not, and whether client work is allowed.&lt;/p&gt;

&lt;h2&gt;
  
  
  7. Can you see a block working before you buy?
&lt;/h2&gt;

&lt;p&gt;A static screenshot hides a lot: how a block behaves at narrow widths, in dark mode, in your theme. You should be able to check all three before paying.&lt;/p&gt;

&lt;h2&gt;
  
  
  8. Is accessibility built in, on every primitive?
&lt;/h2&gt;

&lt;p&gt;Keyboard, focus and screen-reader support are hard to add after the fact, especially on grids and boards. Ask two things: what primitives is it built on, and do the custom blocks inherit the same keyboard and ARIA behaviour? Then test one with a keyboard.&lt;/p&gt;

&lt;h2&gt;
  
  
  9. How do you get updates?
&lt;/h2&gt;

&lt;p&gt;Copied source is yours, and so is keeping it current. Nothing updates on its own. Find out how you pull a newer version of a block, and how you'll see what changed before you overwrite your own edits.&lt;/p&gt;

&lt;h2&gt;
  
  
  What you take on with any copy-paste library
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;The code lives in your repo, so merging fixes after install is part of the job.&lt;/li&gt;
&lt;li&gt;An agent works from the context it has. A registry or MCP server gives it the real APIs.&lt;/li&gt;
&lt;li&gt;Free and paid parts can follow different release schedules. It's worth knowing both.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The checklist
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Question&lt;/th&gt;
&lt;th&gt;What to look for&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Extends or replaces shadcn/ui?&lt;/td&gt;
&lt;td&gt;Source copied into your repo, your theme tokens&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Base UI, Radix UI or React Aria?&lt;/td&gt;
&lt;td&gt;Works with the one you already use&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Agent-ready?&lt;/td&gt;
&lt;td&gt;A registry plus an MCP server with real APIs&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Hard components?&lt;/td&gt;
&lt;td&gt;Data grid, filters, kanban, calendar, gantt&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;What's free?&lt;/td&gt;
&lt;td&gt;The actual licence, MIT or not&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Paid terms?&lt;/td&gt;
&lt;td&gt;One-time or subscription, per-project fees, client work&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Preview before buying?&lt;/td&gt;
&lt;td&gt;Live, responsive, dark mode, your theme&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Accessibility?&lt;/td&gt;
&lt;td&gt;Keyboard, focus and screen-reader support on every primitive&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Updates?&lt;/td&gt;
&lt;td&gt;A clear way to pull new versions without losing your edits&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;No library wins every row for every project. Pick the one that answers the rows you care about.&lt;/p&gt;

&lt;h2&gt;
  
  
  How ReUI answers these
&lt;/h2&gt;

&lt;p&gt;Quoted from reui.io, so you can check each line yourself.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;1. Extends shadcn/ui&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;The CLI copies the source files into your project, same as stock shadcn/ui, so the code lives in your repo and you can edit it freely.&lt;/p&gt;

&lt;p&gt;Every component reads from your shadcn theme tokens (--background, --foreground, --primary, --radius, font family).&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;2. Primitives&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;The same block for either interaction model, so your existing shadcn setup is the one it drops into.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;3. Coding agents&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;The ReUI MCP server connects your coding agent to the live ReUI registry.&lt;/p&gt;

&lt;p&gt;It is free with a ReUI account: the first connection opens a "Sign in with ReUI" approval in your browser, and a free account is created in that flow if you do not have one.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;a href="https://reui.io/docs/mcp?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=shadcn-buyers-guide" rel="noopener noreferrer"&gt;Setup&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;4. Hard components&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Data Grid, Filters, Kanban, Calendar, Gantt and more, headless and yours.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;5. Free licence&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Commercial and open-source use are both fine; only the premium blocks, icons, and paid templates need a license, and templates marked Free need only a free account, under the License on Personal terms rather than MIT.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;6. Paid terms&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;ReUI is a one-time payment with lifetime updates.&lt;/p&gt;

&lt;p&gt;Ship it in as many client and commercial projects as you like.&lt;/p&gt;

&lt;p&gt;One purchase, no per-project fee.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;a href="https://reui.io/pricing?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=shadcn-buyers-guide" rel="noopener noreferrer"&gt;Pricing&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;7. Preview&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Pro blocks give you the full live preview, responsive, dark-mode-toggleable and theme-customizable, so you can confirm one fits your design before buying.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;8. Accessibility&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Accessibility, keyboard, and ARIA baked in&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;9. Updates&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Re-run the install command to pull the latest version of any block.&lt;/p&gt;
&lt;/blockquote&gt;

</description>
      <category>shadcn</category>
      <category>react</category>
      <category>tailwindcss</category>
      <category>webdev</category>
    </item>
    <item>
      <title>How to Use the ReUI shadcn Registry and MCP with Claude Code</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Thu, 05 Nov 2026 06:03:15 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/how-to-use-the-reui-shadcn-registry-and-mcp-with-claude-code-25d</link>
      <guid>https://dev.to/faizalsukaimi/how-to-use-the-reui-shadcn-registry-and-mcp-with-claude-code-25d</guid>
      <description>&lt;p&gt;The ReUI MCP server connects your coding agent to the live ReUI registry. Through one endpoint your agent searches the full library with scored, intent-based matching, reads real component APIs, and gets exact shadcn CLI install commands - so it ships real ReUI instead of guessing.&lt;/p&gt;

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

&lt;p&gt;ReUI is built on the latest &lt;strong&gt;React 19&lt;/strong&gt; and &lt;a href="https://tailwindcss.com" rel="noopener noreferrer"&gt;Tailwind CSS v4&lt;/a&gt;. Ensure your project is set up with these technologies before proceeding.&lt;/p&gt;

&lt;p&gt;We recommend following the &lt;a href="https://ui.shadcn.com/docs/installation" rel="noopener noreferrer"&gt;Official shadcn/ui Installation Guide&lt;/a&gt; and &lt;a href="https://ui.shadcn.com/create" rel="noopener noreferrer"&gt;shadcn/create&lt;/a&gt; setup to prepare your environment.&lt;/p&gt;

&lt;h2&gt;
  
  
  Registry
&lt;/h2&gt;

&lt;p&gt;The ReUI Registry serves every ReUI item from a single &lt;code&gt;@reui&lt;/code&gt; namespace: free primitives and &lt;code&gt;c-*&lt;/code&gt; components, paid blocks and paid icons. The shadcn CLI resolves the full alias suffix into the registry URL, so ReUI items install as &lt;code&gt;@reui/&amp;lt;name&amp;gt;&lt;/code&gt; and ReUI handles access rules on the server.&lt;/p&gt;

&lt;p&gt;Add the ReUI registry namespace to your &lt;code&gt;components.json&lt;/code&gt;.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"registries"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="nl"&gt;"@reui"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"https://reui.io/r/{style}/{name}.json"&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&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;p&gt;There is no ReUI npm package to install. Running &lt;code&gt;npm i @reui/data-grid&lt;/code&gt; asks npm's own &lt;code&gt;@reui&lt;/code&gt; scope, which belongs to an unrelated publisher, so it either fails or installs a package that is not ours.&lt;/p&gt;

&lt;h3&gt;
  
  
  Free Components
&lt;/h3&gt;

&lt;p&gt;Free component examples use the &lt;code&gt;c-*&lt;/code&gt; naming pattern and can be installed without a license key.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;npx shadcn@latest add @reui/c-alert-1
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Claude Code (CLI)
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Add the ReUI MCP server
&lt;/h3&gt;

&lt;p&gt;From your project root, register the server over HTTP:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;claude mcp add &lt;span class="nt"&gt;--transport&lt;/span&gt; http reui https://mcp.reui.io
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The ReUI installer writes the same server into &lt;code&gt;.mcp.json&lt;/code&gt;, spelling the URL out in full as &lt;code&gt;https://mcp.reui.io/api/mcp&lt;/code&gt;. The short form above rewrites to it, so both resolve to the same server and you can use whichever you prefer.&lt;/p&gt;

&lt;h3&gt;
  
  
  Sign in with ReUI
&lt;/h3&gt;

&lt;p&gt;Open the MCP menu inside Claude Code:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;/mcp
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Select &lt;code&gt;reui&lt;/code&gt; and follow the steps in your browser to approve the "Sign in with ReUI" prompt. A free ReUI account is created in that flow if you do not have one.&lt;/p&gt;

&lt;h3&gt;
  
  
  Confirm the connection
&lt;/h3&gt;

&lt;p&gt;Check that Claude Code sees the server and its tools:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;claude mcp list
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;reui&lt;/code&gt; should report &lt;code&gt;✔ Connected&lt;/code&gt;. &lt;code&gt;! Needs authentication&lt;/code&gt; means the sign-in did not finish, and &lt;code&gt;✘ Failed to connect&lt;/code&gt; means Claude Code could not reach the server at all.&lt;/p&gt;

&lt;h3&gt;
  
  
  Install the ReUI skill
&lt;/h3&gt;

&lt;p&gt;This drops the ReUI skill into &lt;code&gt;.claude/skills/reui&lt;/code&gt; so Claude knows how to search the registry, read the real component APIs, and wire blocks together the ReUI way instead of guessing. The same run adds the &lt;code&gt;reui&lt;/code&gt; server to &lt;code&gt;.mcp.json&lt;/code&gt;, adds ReUI's design rules to &lt;code&gt;AGENTS.md&lt;/code&gt; (never created beside a &lt;code&gt;CLAUDE.md&lt;/code&gt;), and writes the skill and MCP entry of the other agents it supports, including a &lt;code&gt;reui&lt;/code&gt; entry in &lt;code&gt;~/.codex/config.toml&lt;/code&gt; in your home directory. Add &lt;code&gt;REUI_DRY_RUN=1&lt;/code&gt; before &lt;code&gt;node&lt;/code&gt; to see every planned write first.&lt;/p&gt;

&lt;p&gt;Run the one-line installer from your project root so Claude Code follows the ReUI build workflow:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl &lt;span class="nt"&gt;-fsSL&lt;/span&gt; https://mcp.reui.io/install | node -
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Build with ReUI
&lt;/h2&gt;

&lt;p&gt;Describe the interface you want in plain language and Claude pulls the right ReUI components, examples, and blocks from the registry, then wires them into your project.&lt;/p&gt;

&lt;p&gt;A free ReUI account covers the free catalog in full - the 27 ReUI components and every &lt;code&gt;c-*&lt;/code&gt; example - across every tool, from &lt;code&gt;search&lt;/code&gt; and &lt;code&gt;compose_page&lt;/code&gt; to the install command for each one.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Add a kanban board for support tickets with drag and drop, grouped by status, and a filters bar above it.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;





&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Add a data-grid of orders with sorting, column filters, pagination, and row selection, wired to my /api/orders endpoint.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Unlock premium items
&lt;/h2&gt;

&lt;p&gt;Premium blocks and icons use the same &lt;code&gt;@reui&lt;/code&gt; namespace, but they require a valid license key.&lt;/p&gt;

&lt;p&gt;Free items (the 27 components and every &lt;code&gt;c-*&lt;/code&gt; example) install with no license. To let Claude install premium blocks and icons too, grab a &lt;a href="https://reui.io/pricing?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=shadcn-registry-tutorial" rel="noopener noreferrer"&gt;Pro or Ultimate license&lt;/a&gt; - one license unlocks the whole premium registry and removes the daily MCP request limit.&lt;/p&gt;

&lt;p&gt;Copy your key from &lt;a href="https://reui.io/account/licenses?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=shadcn-registry-tutorial" rel="noopener noreferrer"&gt;Account → Licenses&lt;/a&gt; into &lt;code&gt;.env.local&lt;/code&gt; at the project root:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="nv"&gt;REUI_LICENSE_KEY&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;your-license-key
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Change the registry entry to the authenticated object form.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"registries"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="nl"&gt;"@reui"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"url"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"https://reui.io/r/{style}/{name}.json"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"headers"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
        &lt;/span&gt;&lt;span class="nl"&gt;"Authorization"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"Bearer ${REUI_LICENSE_KEY}"&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&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;p&gt;Free items keep working with the authenticated config too, so you do not need a second registry namespace.&lt;/p&gt;

&lt;h2&gt;
  
  
  Plans and limits
&lt;/h2&gt;

&lt;p&gt;Free accounts get a daily request allowance, shown on your ReUI account page. Your exact figure and today's usage are on &lt;a href="https://reui.io/account/mcp?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=shadcn-registry-tutorial" rel="noopener noreferrer"&gt;Account → MCP&lt;/a&gt;, which is the only place either number is right for your account.&lt;/p&gt;

&lt;h2&gt;
  
  
  Troubleshooting
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Tools missing, greyed out, or "not connected"&lt;/strong&gt; - the server did not finish connecting. To fix it, run &lt;code&gt;/mcp&lt;/code&gt; in Claude Code and re-authenticate &lt;code&gt;reui&lt;/code&gt; (or toggle the connector off and on in Claude Desktop), then restart Claude.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Sign-in appears to work but every call still 401s&lt;/strong&gt; - check whether you have an &lt;code&gt;Authorization&lt;/code&gt; header on the &lt;code&gt;reui&lt;/code&gt; entry in &lt;code&gt;.mcp.json&lt;/code&gt;. A configured credential is sent on every request and overrides the OAuth credential the sign-in stores, so while it is there running &lt;code&gt;/mcp&lt;/code&gt; and re-authenticating &lt;code&gt;reui&lt;/code&gt; cannot take effect. The two are alternatives, not layers: use the header only on the headless path, and remove it if you want the browser sign-in.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;"Your free ReUI MCP allowance ... is used up" in a tool result&lt;/strong&gt; - the free plan has a daily MCP request allowance and today's is used up. It arrives as a normal tool result Claude shows you, not an HTTP error, so there is no status code to look for, and it resets at UTC midnight.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;a href="https://reui.io/docs/mcp?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=shadcn-registry-tutorial" rel="noopener noreferrer"&gt;Set up ReUI MCP&lt;/a&gt;&lt;/p&gt;

</description>
      <category>shadcn</category>
      <category>mcp</category>
      <category>react</category>
      <category>ai</category>
    </item>
    <item>
      <title>TanStack Table vs ReUI: picking a React data grid for a shadcn app</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Mon, 02 Nov 2026 17:18:53 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/tanstack-table-vs-reui-picking-a-react-data-grid-for-a-shadcn-app-21c1</link>
      <guid>https://dev.to/faizalsukaimi/tanstack-table-vs-reui-picking-a-react-data-grid-for-a-shadcn-app-21c1</guid>
      <description>&lt;p&gt;If you build a data grid with TanStack Table in a shadcn app, most of the work is the grid UI, not the table logic. ReUI's data grid is that UI already written, on top of the same TanStack Table. So the useful question isn't which engine to use. It's how much grid UI you want to write, and how much of it you want to own.&lt;/p&gt;

&lt;p&gt;I work on ReUI, so keep that in mind. I'll point out where plain TanStack Table is the better call.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I checked
&lt;/h2&gt;

&lt;p&gt;I compared &lt;code&gt;@tanstack/react-table&lt;/code&gt; 9.3.1 with the ReUI data grid as the shadcn CLI served it on 3 Nov 2026. To see exactly what lands in a project, I ran &lt;code&gt;npx shadcn@latest view @reui/data-grid&lt;/code&gt; and counted the files.&lt;/p&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;TanStack Table&lt;/th&gt;
&lt;th&gt;ReUI data grid&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;What you install&lt;/td&gt;
&lt;td&gt;one npm package&lt;/td&gt;
&lt;td&gt;12 files in &lt;code&gt;components/reui/data-grid&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;UI included&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;td&gt;table, header menus, filters, pagination, virtual table, cell selection&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Extra packages&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;td&gt;TanStack Virtual, dnd-kit, Base UI or Radix&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Who maintains the UI&lt;/td&gt;
&lt;td&gt;you&lt;/td&gt;
&lt;td&gt;you, once it's installed&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;License&lt;/td&gt;
&lt;td&gt;MIT&lt;/td&gt;
&lt;td&gt;MIT, free&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  TanStack Table gives you state, not a table
&lt;/h2&gt;

&lt;p&gt;TanStack Table describes itself as headless, with "full control over markup, styles, and behavior". You get sorting, filtering, grouping, selection and pagination as state and row models. It renders nothing.&lt;/p&gt;

&lt;p&gt;For a sortable, paged table that's ideal. You map rows into the shadcn &lt;code&gt;&amp;lt;Table&amp;gt;&lt;/code&gt; you already have and you're done in an afternoon.&lt;/p&gt;

&lt;p&gt;It gets heavier when the table has to behave like a grid: a menu on each column header, resize handles, pinned columns, keyboard focus moving between cells, virtualization (a separate package, TanStack Virtual), and copy and paste across a range of cells. TanStack Table leaves all of that UI to you on purpose.&lt;/p&gt;

&lt;h2&gt;
  
  
  ReUI's data grid is that UI, already written
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;npx shadcn@latest add @reui/data-grid
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;That adds 12 files and a little over 11,000 lines of TSX to your project. The four biggest files:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;File&lt;/th&gt;
&lt;th&gt;Lines (about)&lt;/th&gt;
&lt;th&gt;What it covers&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;data-grid-table.tsx&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;3,100&lt;/td&gt;
&lt;td&gt;the table, sticky headers, pinning, row states&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;data-grid-cell-selection.tsx&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;2,600&lt;/td&gt;
&lt;td&gt;range selection, copy, paste, fill handle&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;data-grid.tsx&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;1,300&lt;/td&gt;
&lt;td&gt;the provider and shared state&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;data-grid-table-virtual.tsx&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;1,200&lt;/td&gt;
&lt;td&gt;row and column virtualization&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;That number cuts both ways. It's a lot of grid work you skip. It's also a lot of code that now lives in your repo: you can change any line, and you also maintain every line.&lt;/p&gt;

&lt;p&gt;One caveat on the cell selection file: a range selection you write yourself could be far smaller if you skip the edge cases (pinned columns, virtualized rows, translated labels). The count shows what a complete version covers, not what you'd have to write.&lt;/p&gt;

&lt;p&gt;You still write the TanStack Table part. The grid renders it:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight tsx"&gt;&lt;code&gt;&lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="nx"&gt;table&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;useTable&lt;/span&gt;&lt;span class="p"&gt;({&lt;/span&gt;
  &lt;span class="na"&gt;features&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nx"&gt;dataGridFeatures&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="nx"&gt;data&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="nx"&gt;columns&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="p"&gt;})&lt;/span&gt;

&lt;span class="k"&gt;return &lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
  &lt;span class="p"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nc"&gt;DataGrid&lt;/span&gt; &lt;span class="na"&gt;table&lt;/span&gt;&lt;span class="p"&gt;=&lt;/span&gt;&lt;span class="si"&gt;{&lt;/span&gt;&lt;span class="nx"&gt;table&lt;/span&gt;&lt;span class="si"&gt;}&lt;/span&gt; &lt;span class="na"&gt;recordCount&lt;/span&gt;&lt;span class="p"&gt;=&lt;/span&gt;&lt;span class="si"&gt;{&lt;/span&gt;&lt;span class="nx"&gt;data&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;length&lt;/span&gt;&lt;span class="si"&gt;}&lt;/span&gt;&lt;span class="p"&gt;&amp;gt;&lt;/span&gt;
    &lt;span class="p"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nc"&gt;DataGridContainer&lt;/span&gt;&lt;span class="p"&gt;&amp;gt;&lt;/span&gt;
      &lt;span class="p"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nc"&gt;DataGridTable&lt;/span&gt; &lt;span class="p"&gt;/&amp;gt;&lt;/span&gt;
    &lt;span class="p"&gt;&amp;lt;/&lt;/span&gt;&lt;span class="nc"&gt;DataGridContainer&lt;/span&gt;&lt;span class="p"&gt;&amp;gt;&lt;/span&gt;
    &lt;span class="p"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nc"&gt;DataGridPagination&lt;/span&gt; &lt;span class="p"&gt;/&amp;gt;&lt;/span&gt;
  &lt;span class="p"&gt;&amp;lt;/&lt;/span&gt;&lt;span class="nc"&gt;DataGrid&lt;/span&gt;&lt;span class="p"&gt;&amp;gt;&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Your columns, row models and state work the way they do in TanStack Table. Grid behaviour is turned on with &lt;code&gt;tableLayout&lt;/code&gt; props instead of built by hand: &lt;code&gt;headerSticky&lt;/code&gt;, &lt;code&gt;columnsResizable&lt;/code&gt;, &lt;code&gt;columnsPinnable&lt;/code&gt;, &lt;code&gt;columnsDraggable&lt;/code&gt;, &lt;code&gt;rowsPinnable&lt;/code&gt;, &lt;code&gt;cellSelection&lt;/code&gt;. On top of that you get infinite scroll, tree and expandable rows, footer rows, inline editing, and one &lt;code&gt;i18n&lt;/code&gt; prop for every built-in string.&lt;/p&gt;

&lt;p&gt;Four of the free examples:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/components/data-grid/c-data-grid-31?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=data-grid-comparison" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ff108qsrglq0vbh8x0xnm.webp" alt="Spreadsheet editing" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/components/data-grid/c-data-grid-32?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=data-grid-comparison" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7mfho0xlhz15h23ljcyj.webp" alt="Column virtualization" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/components/data-grid/c-data-grid-18?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=data-grid-comparison" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F33gh8xkofbqn96zy3zzi.webp" alt="Column controls" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/components/data-grid/c-data-grid-30?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=data-grid-comparison" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fqo2xzmym7uuhydcyjpgx.webp" alt="Tree rows" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;There are two versions, one on Base UI and one on Radix UI, so use the docs page that matches your project. What it doesn't have: pivot tables and Excel export.&lt;/p&gt;

&lt;h2&gt;
  
  
  Which one I'd pick
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;You need&lt;/th&gt;
&lt;th&gt;Pick&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;A sortable, paged table&lt;/td&gt;
&lt;td&gt;TanStack Table with the shadcn &lt;code&gt;&amp;lt;Table&amp;gt;&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Every pixel your own, and time to build it&lt;/td&gt;
&lt;td&gt;TanStack Table&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Pinning, resizing and virtualization in your own design system&lt;/td&gt;
&lt;td&gt;ReUI data grid&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Spreadsheet-style range selection and editing&lt;/td&gt;
&lt;td&gt;ReUI data grid&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Docs and every example: &lt;a href="https://reui.io/docs/components/base/data-grid?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=data-grid-comparison" rel="noopener noreferrer"&gt;https://reui.io/docs/components/base/data-grid?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=data-grid-comparison&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;If you've built a grid on TanStack Table, what took the longest: the column menus, virtualization, range selection, or something I didn't list?&lt;/p&gt;

</description>
      <category>react</category>
      <category>webdev</category>
      <category>javascript</category>
      <category>frontend</category>
    </item>
    <item>
      <title>Making a component library agent-ready: the ReUI MCP server</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Wed, 28 Oct 2026 09:14:37 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/making-a-component-library-agent-ready-the-reui-mcp-server-3a9i</link>
      <guid>https://dev.to/faizalsukaimi/making-a-component-library-agent-ready-the-reui-mcp-server-3a9i</guid>
      <description>&lt;p&gt;Coding agents are good at writing React. They are bad at knowing your component library. Ask one to "add a data grid with sorting" and it will happily invent a &lt;code&gt;sortable&lt;/code&gt; prop that has never existed, or install a package name it made up. The code looks right and fails on the first render.&lt;/p&gt;

&lt;p&gt;We hit this with ReUI, our set of components and blocks for shadcn/ui, so we built an MCP server that gives agents the real registry instead of their memory of it.&lt;/p&gt;

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

&lt;p&gt;The ReUI MCP server runs at &lt;code&gt;https://mcp.reui.io&lt;/code&gt; over Streamable HTTP, so any MCP-capable agent can use it. It does discovery and guidance only:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Find:&lt;/strong&gt; search components, examples, blocks and icons in plain language.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Read:&lt;/strong&gt; return each component's real API inline, so the agent never has to scrape a docs page.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Plan and validate:&lt;/strong&gt; plan a whole page from an intent, and check planned props before any code is written.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Install:&lt;/strong&gt; return the exact shadcn CLI command for an item, after checking that it exists.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;It never serves source code. Installation stays with the shadcn CLI, which is also where a license key lives for premium items.&lt;/p&gt;

&lt;h2&gt;
  
  
  Catching invented props before they ship
&lt;/h2&gt;

&lt;p&gt;The tool I like most is &lt;code&gt;validate_usage&lt;/code&gt;. The agent sends the component and props it plans to use, and gets a verdict per prop, checked against the indexed API docs. No model, no guessing.&lt;/p&gt;

&lt;p&gt;Here is a real call. The agent plans a data grid with &lt;code&gt;table&lt;/code&gt;, &lt;code&gt;recordCount&lt;/code&gt;, &lt;code&gt;isLoading&lt;/code&gt; and &lt;code&gt;sortable&lt;/code&gt;, and wants to install &lt;code&gt;c-data-grid-12&lt;/code&gt; and &lt;code&gt;data-grid-pro&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"ok"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;false&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"components"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;[{&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="nl"&gt;"name"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"data-grid"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="nl"&gt;"exists"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="nl"&gt;"props"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"prop"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"table"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"documented"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;},&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"prop"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"recordCount"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"documented"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;},&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"prop"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"isLoading"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"documented"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;},&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"prop"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"sortable"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"documented"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;false&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"near"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"table"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"sortingStrategy"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="p"&gt;}],&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"items"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"name"&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-data-grid-12"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"exists"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"type"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"example"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"free"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;},&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"name"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"data-grid-pro"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"exists"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;false&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"suggestions"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;[{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"name"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"data-grid-advanced-7"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"type"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"block"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;},&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"name"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"data-grid"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"type"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"component"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}]&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="w"&gt;
&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;p&gt;&lt;code&gt;sortable&lt;/code&gt; was the invented prop, and &lt;code&gt;data-grid-pro&lt;/code&gt; was the invented package. Both are caught before a single file changes, with a nudge toward what actually exists.&lt;/p&gt;

&lt;h2&gt;
  
  
  Planning a page, not just a component
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;compose_page&lt;/code&gt; takes an intent like "SaaS billing settings page" and returns an ordered plan: an app shell, a settings block, a form, a dialog, each with a top pick, alternates, preview links and the install sequence. Pass &lt;code&gt;surface: "card"&lt;/code&gt; and every section comes from the same design-system surface, so the page doesn't end up as a patchwork of styles.&lt;/p&gt;

&lt;p&gt;The four picks from that run:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/preview/base/app-shell-22?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=mcp-article" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdtxipboummhzvz2qrms3.webp" alt="app-shell-22" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/preview/base/settings-7?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=mcp-article" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fxb5qiigp8d7t4j04npfk.webp" alt="settings-7" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/preview/base/form-5?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=mcp-article" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjrhu12m6oj22mwdmadi4.webp" alt="form-5" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://reui.io/preview/base/dialog-10?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=mcp-article" rel="noopener noreferrer"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fgu318sner5noa1cm7mvx.webp" alt="dialog-10" width="800" height="533"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;If a section has no good match, it is listed under &lt;code&gt;unavailableSections&lt;/code&gt; instead of being filled with the closest wrong thing. That rule matters more than any ranking: an agent that admits a gap is easier to work with than one that papers over it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Setup
&lt;/h2&gt;

&lt;p&gt;Claude Code, from your project root:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;claude mcp add &lt;span class="nt"&gt;--transport&lt;/span&gt; http reui https://mcp.reui.io
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Then run &lt;code&gt;/mcp&lt;/code&gt;, pick &lt;code&gt;reui&lt;/code&gt;, and approve the sign-in in your browser.&lt;/p&gt;

&lt;p&gt;Cursor, in &lt;code&gt;.cursor/mcp.json&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"mcpServers"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="nl"&gt;"reui"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"url"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"https://mcp.reui.io"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&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;p&gt;The first tool call opens a browser to sign in; a free account is created if you don't have one. For CI, create a personal token at Account → MCP and send it as &lt;code&gt;Authorization: Bearer reui_pat_...&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;A free account covers all the free components and &lt;code&gt;c-*&lt;/code&gt; examples. A Pro license adds the premium blocks to search, page plans and installs.&lt;/p&gt;

&lt;h2&gt;
  
  
  Pair it with the skill
&lt;/h2&gt;

&lt;p&gt;The server answers questions. The ReUI Agent Skill tells the agent which questions to ask, and in what order: plan, find, install, read the API, adapt by reuse, then check. Install it with:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl &lt;span class="nt"&gt;-fsSL&lt;/span&gt; https://mcp.reui.io/install | node -
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;In Claude Code it lands in &lt;code&gt;.claude/skills/reui&lt;/code&gt; and adds &lt;code&gt;/mcp__reui__build&lt;/code&gt;, &lt;code&gt;/mcp__reui__add&lt;/code&gt;, &lt;code&gt;/mcp__reui__fix&lt;/code&gt; and &lt;code&gt;/mcp__reui__improve&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;Docs and setup for other agents (VS Code, Gemini CLI and more): &lt;a href="https://reui.io/docs/mcp?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=mcp-article" rel="noopener noreferrer"&gt;https://reui.io/docs/mcp?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=mcp-article&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;I'd like to hear where it falls short for you, especially props it still lets through.&lt;/p&gt;

</description>
      <category>mcp</category>
      <category>ai</category>
      <category>react</category>
      <category>webdev</category>
    </item>
    <item>
      <title>I Built an AI-Powered Database Schema Explorer for the H0 Hackathon</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Mon, 29 Jun 2026 11:53:28 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/i-built-an-ai-powered-database-schema-explorer-for-the-h0-hackathon-3he3</link>
      <guid>https://dev.to/faizalsukaimi/i-built-an-ai-powered-database-schema-explorer-for-the-h0-hackathon-3he3</guid>
      <description>&lt;p&gt;&lt;em&gt;The following content was created for the purposes of entering the H0 Hackathon. #H0Hackathon&lt;/em&gt;&lt;/p&gt;




&lt;p&gt;You join a new project. The database has 30 tables. No documentation. No ERD. No one remembers why &lt;code&gt;status_v2_legacy&lt;/code&gt; exists.&lt;/p&gt;

&lt;p&gt;Existing tools force a tradeoff. pgAdmin gives raw access but no visualization. dbdiagram.io lets you draw schemas but requires manual input that goes stale after every migration. SchemaSpy needs CLI setup and Java. DBeaver is desktop-only — you can't share a link with a new hire.&lt;/p&gt;

&lt;p&gt;I looked for a web tool that auto-discovers schemas, visualizes them, documents them, and lets you query them with AI. I found nothing that does all four.&lt;/p&gt;

&lt;p&gt;So I built &lt;strong&gt;SchemaLens&lt;/strong&gt; for the H0 Hackathon.&lt;/p&gt;

&lt;h2&gt;
  
  
  What SchemaLens Does
&lt;/h2&gt;

&lt;p&gt;Paste a PostgreSQL connection string. That's it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;SchemaLens&lt;/strong&gt; queries &lt;code&gt;information_schema&lt;/code&gt; to pull tables, columns, types, constraints, foreign keys, and row estimates. No manual input. No DSL. No install.&lt;/p&gt;

&lt;p&gt;From there, you get:&lt;/p&gt;

&lt;h3&gt;
  
  
  1. Interactive ERD
&lt;/h3&gt;

&lt;p&gt;A full entity-relationship diagram rendered automatically from your real schema. Zoom, pan, click any table to see columns with types, nullability, defaults, and constraints. Foreign keys become edges. The diagram builds itself.&lt;/p&gt;

&lt;h3&gt;
  
  
  2. Schema Health Scoring
&lt;/h3&gt;

&lt;p&gt;10 automated checks produce a weighted score with letter grade (A–F):&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Primary keys present&lt;/li&gt;
&lt;li&gt;Foreign key validity&lt;/li&gt;
&lt;li&gt;Naming conventions&lt;/li&gt;
&lt;li&gt;Column types&lt;/li&gt;
&lt;li&gt;NOT NULL constraints&lt;/li&gt;
&lt;li&gt;Timestamps&lt;/li&gt;
&lt;li&gt;Oversized columns&lt;/li&gt;
&lt;li&gt;Unique constraints&lt;/li&gt;
&lt;li&gt;Boolean naming&lt;/li&gt;
&lt;li&gt;Column count&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Each finding includes a severity level and suggested fix.&lt;/p&gt;

&lt;h3&gt;
  
  
  3. AI Schema Agent
&lt;/h3&gt;

&lt;p&gt;This is the feature I'm most proud of. An AI assistant that understands your schema and executes queries against your live database.&lt;/p&gt;

&lt;p&gt;Ask in plain English:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;"Show me all users who signed up last month"&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;The agent writes the query, validates it, runs it against your actual database, shows results in a table, and auto-retries on errors.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The tool-calling pipeline:&lt;/strong&gt;&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;generate_sql → check_sql → execute_sql → retry loop
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;You see the agent's reasoning, tool calls, and results streaming in real-time.&lt;/p&gt;

&lt;h3&gt;
  
  
  4. Shareable Links
&lt;/h3&gt;

&lt;p&gt;One click generates a public URL. Your team sees the full ERD and docs without signing up.&lt;/p&gt;

&lt;h2&gt;
  
  
  The Security Model
&lt;/h2&gt;

&lt;p&gt;This was the hardest part. The AI agent generates and executes SQL against a live database. That's terrifying.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Two-layer defense:&lt;/strong&gt;&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Application layer&lt;/strong&gt; — blocks destructive keywords (&lt;code&gt;INSERT&lt;/code&gt;, &lt;code&gt;UPDATE&lt;/code&gt;, &lt;code&gt;DELETE&lt;/code&gt;, &lt;code&gt;DROP&lt;/code&gt;, &lt;code&gt;ALTER&lt;/code&gt;, &lt;code&gt;TRUNCATE&lt;/code&gt;) and injection attempts via regex&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Database layer&lt;/strong&gt; — PostgreSQL itself is set to &lt;code&gt;READ ONLY&lt;/code&gt; mode, rejecting any write regardless of what the application misses&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Even if the AI generates an &lt;code&gt;INSERT&lt;/code&gt;, PostgreSQL rejects it. The agent cannot write data.&lt;/p&gt;

&lt;p&gt;This is defense in depth, not just a prompt telling the AI to be nice.&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Layer&lt;/th&gt;
&lt;th&gt;Technology&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Framework&lt;/td&gt;
&lt;td&gt;Next.js 16, React, TypeScript&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;UI&lt;/td&gt;
&lt;td&gt;Tailwind v4, shadcn/ui, React Flow&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Auth&lt;/td&gt;
&lt;td&gt;better-auth (email/password)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ORM&lt;/td&gt;
&lt;td&gt;Drizzle ORM&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Database&lt;/td&gt;
&lt;td&gt;Amazon Aurora PostgreSQL 16&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;LLM&lt;/td&gt;
&lt;td&gt;OpenAI-compatible (DeepSeek V4 Flash)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AI SDK&lt;/td&gt;
&lt;td&gt;Vercel AI SDK — streamText + tool calling&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Deployment&lt;/td&gt;
&lt;td&gt;Vercel&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Aurora PostgreSQL&lt;/strong&gt; is doing the heavy lifting:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Auto-discovers schemas from the system catalog&lt;/li&gt;
&lt;li&gt;Stores schema snapshots as JSONB&lt;/li&gt;
&lt;li&gt;Handles user data, sessions, chat history, and conversations&lt;/li&gt;
&lt;li&gt;Serverless scaling — no always-on servers&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The Vercel + Aurora combo is legit. Not just a hackathon shortcut — clean integration, fast deploys, real scaling potential.&lt;/p&gt;

&lt;h2&gt;
  
  
  Challenges I Hit
&lt;/h2&gt;

&lt;h3&gt;
  
  
  SSL for Aurora on Vercel
&lt;/h3&gt;

&lt;p&gt;Aurora requires SSL but the database driver rejects self-signed certs. Connection pool failed silently until I configured it to accept the certificate chain. Silent failures are the worst.&lt;/p&gt;

&lt;h3&gt;
  
  
  Better-auth Drizzle Adapter
&lt;/h3&gt;

&lt;p&gt;Table key naming mismatch caused silent auth failures. The handler swallowed errors and returned empty responses. Had to debug the handler output to find the real error.&lt;/p&gt;

&lt;h3&gt;
  
  
  AI Agent SQL Safety
&lt;/h3&gt;

&lt;p&gt;Getting the self-correcting retry loop to work reliably with real-time streaming took significant effort. When the AI generates invalid SQL, it reads the database error, fixes the query, and retries automatically.&lt;/p&gt;

&lt;h3&gt;
  
  
  Schema Health Tuning
&lt;/h3&gt;

&lt;p&gt;What looks good on paper (check every table for primary keys) produces noise on real schemas. Each check needed iteration to balance sensitivity with specificity.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Learned
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;PostgreSQL's system catalog is rich&lt;/strong&gt; — the database already knows everything about itself. You just have to ask.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;AI agents need guardrails, not just prompts&lt;/strong&gt; — the schema chat works because of defense in depth.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Self-correcting agents need boundaries&lt;/strong&gt; — the generate → validate → execute → retry loop lets the agent fix its own mistakes, but you need clear retry limits.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Vercel + Aurora is production-ready&lt;/strong&gt; — not just a hackathon shortcut.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Try It Yourself
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Go to &lt;a href="https://schemalens-alpha.vercel.app" rel="noopener noreferrer"&gt;schemalens-alpha.vercel.app&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Sign up for a free account&lt;/li&gt;
&lt;li&gt;Click "Connect database"&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Paste your PostgreSQL connection string (any PostgreSQL DB works — Aurora, Railway, Supabase, etc.)&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;The ERD renders automatically&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;Open &lt;strong&gt;Schema Chat&lt;/strong&gt; and try:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;"list all tables"&lt;/li&gt;
&lt;li&gt;"which tables have no foreign keys?"&lt;/li&gt;
&lt;li&gt;"show me the 5 largest tables by row count"&lt;/li&gt;
&lt;li&gt;"if I delete a connector, which related data would be affected?"&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Click "Share" to generate a public link&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  What's Next
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Schema diff&lt;/strong&gt; — re-introspect to see what changed&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Export&lt;/strong&gt; — SQL migrations, dbdiagram.io DSL, markdown docs&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Team features&lt;/strong&gt; — shared schema libraries, RBAC, Slack integration&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Multi-database&lt;/strong&gt; — MySQL, SQLite, MongoDB&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Schema monitoring&lt;/strong&gt; — scheduled re-introspection with drift alerts&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Source Code
&lt;/h2&gt;

&lt;p&gt;GitHub: &lt;a href="https://github.com/faizalmy/schemalens" rel="noopener noreferrer"&gt;github.com/faizalmy/schemalens&lt;/a&gt;&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Built with ❤️ for the #H0Hackathon. Thanks to AWS Aurora and Vercel for the platforms.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>h0hackathon</category>
      <category>postgres</category>
      <category>ai</category>
      <category>aws</category>
    </item>
    <item>
      <title>Building a Legal Document Intelligence Platform with BigQuery AI: 99% Efficiency Implementation Guide</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Tue, 16 Sep 2025 08:54:44 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/building-a-legal-document-intelligence-platform-with-bigquery-ai-99-efficiency-implementation-5ajo</link>
      <guid>https://dev.to/faizalsukaimi/building-a-legal-document-intelligence-platform-with-bigquery-ai-99-efficiency-implementation-5ajo</guid>
      <description>&lt;h2&gt;
  
  
  Building a Legal Document Intelligence Platform with BigQuery AI: 99% Efficiency Implementation Guide
&lt;/h2&gt;

&lt;p&gt;&lt;em&gt;How I leveraged all 6 BigQuery AI functions to create a legal document processing system that achieves 99%+ efficiency improvements and solves real-world legal industry challenges&lt;/em&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Introduction
&lt;/h3&gt;

&lt;p&gt;Legal professionals spend countless hours manually processing documents, extracting key information, and searching for relevant precedents. What if we could automate this entire workflow using BigQuery's cutting-edge AI capabilities?&lt;/p&gt;

&lt;p&gt;In this post, I'll walk you through my complete implementation of a &lt;strong&gt;Legal Document Intelligence Platform&lt;/strong&gt; that leverages all 6 BigQuery AI functions to transform how legal documents are processed, analyzed, and searched. The results? &lt;strong&gt;99.2% efficiency improvements&lt;/strong&gt; in document summarization and &lt;strong&gt;55.1%-70.0% similarity matching&lt;/strong&gt; in semantic search.&lt;/p&gt;

&lt;h3&gt;
  
  
  The Challenge: Legal Document Processing Pain Points
&lt;/h3&gt;

&lt;p&gt;The legal industry faces several critical challenges:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Manual Processing&lt;/strong&gt;: Lawyers spend hours reading and summarizing lengthy legal documents&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Data Extraction Inefficiency&lt;/strong&gt;: Critical information buried in unstructured text requires manual extraction&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Search Limitations&lt;/strong&gt;: Traditional keyword-based search misses semantically relevant precedents&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scalability Issues&lt;/strong&gt;: Processing thousands of documents becomes increasingly time-consuming&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  The Solution: BigQuery AI Architecture
&lt;/h3&gt;

&lt;p&gt;I designed a comprehensive solution that combines &lt;strong&gt;Generative AI&lt;/strong&gt; and &lt;strong&gt;Vector Search&lt;/strong&gt; to address all these challenges:&lt;/p&gt;

&lt;p&gt;Generative AI (The AI Architect)&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;ML.GENERATE_TEXT&lt;/strong&gt;: Automated legal document summarization&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;AI.GENERATE_TABLE&lt;/strong&gt;: Structured legal data extraction&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;AI.GENERATE_BOOL&lt;/strong&gt;: Urgency detection and classification&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;AI.FORECAST&lt;/strong&gt;: Case outcome prediction and trend analysis&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Vector Search (The Semantic Detective)&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;ML.GENERATE_EMBEDDING&lt;/strong&gt;: Document embeddings for semantic search&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;VECTOR_SEARCH&lt;/strong&gt;: Similarity search and document matching&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;ML.DISTANCE&lt;/strong&gt;: Precise similarity calculations&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  Technical Implementation
&lt;/h3&gt;

&lt;h4&gt;
  
  
  Core BigQuery AI Function Implementation
&lt;/h4&gt;

&lt;p&gt;Here's how I implemented each function with legal domain-specific optimizations:&lt;/p&gt;

&lt;h4&gt;
  
  
  ML.GENERATE_TEXT - Document Summarization
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Purpose&lt;/strong&gt;: Automatically generate concise summaries of lengthy legal documents, extracting key legal issues and outcomes.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How it works&lt;/strong&gt;: Uses Google's Gemini Pro model to analyze the full document content and generate a 3-sentence summary focused on legal relevance. The prompt engineering ensures the AI focuses on legal issues rather than general content.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key Benefits&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Reduces 10-15 page legal documents to 3 focused sentences&lt;/li&gt;
&lt;li&gt;Maintains legal accuracy while improving readability&lt;/li&gt;
&lt;li&gt;Enables quick case assessment and prioritization
&lt;/li&gt;
&lt;/ul&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt;
  &lt;span class="n"&gt;document_id&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;document_type&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;ML&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;GENERATE_TEXT&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;MODEL&lt;/span&gt; &lt;span class="nv"&gt;`project.ai_models.gemini_pro`&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;CONCAT&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
      &lt;span class="s1"&gt;'Summarize this legal document in 3 sentences, focusing on key legal issues and outcomes: '&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="n"&gt;content&lt;/span&gt;
    &lt;span class="p"&gt;)&lt;/span&gt;
  &lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;summary&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;`project.legal_ai_platform_raw_data.legal_documents`&lt;/span&gt;
&lt;span class="k"&gt;WHERE&lt;/span&gt; &lt;span class="n"&gt;document_type&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s1"&gt;'case_law'&lt;/span&gt;
&lt;span class="k"&gt;LIMIT&lt;/span&gt; &lt;span class="mi"&gt;10&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h4&gt;
  
  
  AI.GENERATE_TABLE - Structured Data Extraction
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Purpose&lt;/strong&gt;: Extract structured legal entities from unstructured text, converting complex legal documents into organized, queryable data.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How it works&lt;/strong&gt;: Uses AI to identify and extract specific legal entities (case numbers, parties, dates, amounts) from unstructured text. The STRUCT function defines the output schema, ensuring consistent data format for downstream analysis.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key Benefits&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Transforms unstructured legal text into structured database records&lt;/li&gt;
&lt;li&gt;Enables automated case management and tracking&lt;/li&gt;
&lt;li&gt;Facilitates legal analytics and reporting&lt;/li&gt;
&lt;li&gt;Reduces manual data entry errors
&lt;/li&gt;
&lt;/ul&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt;
  &lt;span class="n"&gt;document_id&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;AI&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;GENERATE_TABLE&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;MODEL&lt;/span&gt; &lt;span class="nv"&gt;`project.ai_models.gemini_pro`&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;CONCAT&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
      &lt;span class="s1"&gt;'Extract the following legal entities as JSON: case_number, court_name, case_date, plaintiff, defendant, monetary_amount, legal_issues. Document: '&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="n"&gt;content&lt;/span&gt;
    &lt;span class="p"&gt;),&lt;/span&gt;
    &lt;span class="n"&gt;STRUCT&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
      &lt;span class="s1"&gt;'case_number'&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;case_number&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="s1"&gt;'court_name'&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;court_name&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="s1"&gt;'case_date'&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;case_date&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="s1"&gt;'plaintiff'&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;plaintiff&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="s1"&gt;'defendant'&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;defendant&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="s1"&gt;'monetary_amount'&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;monetary_amount&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="s1"&gt;'legal_issues'&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;legal_issues&lt;/span&gt;
    &lt;span class="p"&gt;)&lt;/span&gt;
  &lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;extracted_data&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;`project.legal_ai_platform_raw_data.legal_documents`&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h4&gt;
  
  
  AI.GENERATE_BOOL - Urgency Detection
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Purpose&lt;/strong&gt;: Automatically classify legal documents by urgency level, helping legal professionals prioritize their workload and identify time-sensitive matters.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How it works&lt;/strong&gt;: Uses AI to analyze document content for urgency indicators such as deadlines, emergency language, time-sensitive legal matters, or critical case elements. Returns a simple true/false classification for easy filtering and prioritization.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key Benefits&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Automatically flags urgent documents requiring immediate attention&lt;/li&gt;
&lt;li&gt;Enables workload prioritization and deadline management&lt;/li&gt;
&lt;li&gt;Reduces risk of missing critical deadlines&lt;/li&gt;
&lt;li&gt;Improves case management efficiency
&lt;/li&gt;
&lt;/ul&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt;
  &lt;span class="n"&gt;document_id&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;AI&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;GENERATE_BOOL&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;MODEL&lt;/span&gt; &lt;span class="nv"&gt;`project.ai_models.gemini_pro`&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;CONCAT&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
      &lt;span class="s1"&gt;'Is this legal document urgent and requires immediate attention? Respond with true or false. Document: '&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="n"&gt;content&lt;/span&gt;
    &lt;span class="p"&gt;)&lt;/span&gt;
  &lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;is_urgent&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;`project.legal_ai_platform_raw_data.legal_documents`&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h4&gt;
  
  
  AI.FORECAST - Case Outcome Prediction
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Purpose&lt;/strong&gt;: Predict future case outcomes and trends based on historical legal data patterns, enabling strategic planning and resource allocation.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How it works&lt;/strong&gt;: Uses time-series forecasting models to analyze historical case data and predict future outcomes. The model generates forecasts with confidence intervals, providing both predictions and uncertainty measures for decision-making.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key Benefits&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Enables strategic case planning and resource allocation&lt;/li&gt;
&lt;li&gt;Provides data-driven insights for legal strategy&lt;/li&gt;
&lt;li&gt;Helps identify trends and patterns in case outcomes&lt;/li&gt;
&lt;li&gt;Supports risk assessment and client communication
&lt;/li&gt;
&lt;/ul&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt;
  &lt;span class="n"&gt;forecast_timestamp&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;forecast_value&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;confidence_interval_lower_bound&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;confidence_interval_upper_bound&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="n"&gt;ML&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;FORECAST&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
  &lt;span class="n"&gt;MODEL&lt;/span&gt; &lt;span class="nv"&gt;`project.ai_models.legal_timesfm`&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;STRUCT&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mi"&gt;7&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;horizon&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="mi"&gt;95&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;confidence_level&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;h4&gt;
  
  
  ML.GENERATE_EMBEDDING - Document Embeddings
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Purpose&lt;/strong&gt;: Convert legal documents into high-dimensional vector representations that capture semantic meaning, enabling sophisticated similarity search and document clustering.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How it works&lt;/strong&gt;: Uses Google's text embedding model to transform document content into numerical vectors that represent semantic meaning. These embeddings capture the conceptual relationships between legal concepts, enabling AI to understand document similarity beyond keyword matching.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key Benefits&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Enables semantic search that understands legal concepts and relationships&lt;/li&gt;
&lt;li&gt;Powers document clustering and categorization&lt;/li&gt;
&lt;li&gt;Facilitates precedent discovery and case similarity analysis&lt;/li&gt;
&lt;li&gt;Supports advanced legal research and analysis
&lt;/li&gt;
&lt;/ul&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt;
  &lt;span class="n"&gt;document_id&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;ML&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;GENERATE_EMBEDDING&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;MODEL&lt;/span&gt; &lt;span class="nv"&gt;`project.ai_models.text_embedding`&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;content&lt;/span&gt;
  &lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;embedding&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;`project.legal_ai_platform_raw_data.legal_documents`&lt;/span&gt;
&lt;span class="k"&gt;WHERE&lt;/span&gt; &lt;span class="n"&gt;content&lt;/span&gt; &lt;span class="k"&gt;IS&lt;/span&gt; &lt;span class="k"&gt;NOT&lt;/span&gt; &lt;span class="k"&gt;NULL&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h4&gt;
  
  
  VECTOR_SEARCH - Semantic Similarity Search
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Purpose&lt;/strong&gt;: Find semantically similar legal documents based on meaning rather than keywords, enabling intelligent precedent discovery and case relationship analysis.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How it works&lt;/strong&gt;: Compares document embeddings using cosine similarity to find the most relevant legal precedents. The search understands legal concepts and relationships, returning documents that are semantically similar even if they use different terminology.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key Benefits&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Discovers relevant legal precedents based on semantic meaning&lt;/li&gt;
&lt;li&gt;Enables intelligent case research and precedent discovery&lt;/li&gt;
&lt;li&gt;Finds related cases that keyword search might miss&lt;/li&gt;
&lt;li&gt;Supports comprehensive legal analysis and strategy development
&lt;/li&gt;
&lt;/ul&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt;
  &lt;span class="n"&gt;base&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;document_id&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;distance&lt;/span&gt; &lt;span class="k"&gt;AS&lt;/span&gt; &lt;span class="n"&gt;similarity_distance&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="n"&gt;VECTOR_SEARCH&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
  &lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="k"&gt;SELECT&lt;/span&gt; &lt;span class="n"&gt;document_id&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;embedding&lt;/span&gt;
    &lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;`project.legal_ai_platform_vector_indexes.document_embeddings`&lt;/span&gt;
  &lt;span class="p"&gt;),&lt;/span&gt;
  &lt;span class="s1"&gt;'embedding'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="k"&gt;SELECT&lt;/span&gt; &lt;span class="n"&gt;ML&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;GENERATE_EMBEDDING&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
      &lt;span class="n"&gt;MODEL&lt;/span&gt; &lt;span class="nv"&gt;`project.ai_models.text_embedding`&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="s1"&gt;'legal case about marriage'&lt;/span&gt;
    &lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;query_embedding&lt;/span&gt;
  &lt;span class="p"&gt;),&lt;/span&gt;
  &lt;span class="n"&gt;top_k&lt;/span&gt; &lt;span class="o"&gt;=&amp;gt;&lt;/span&gt; &lt;span class="mi"&gt;5&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;distance_type&lt;/span&gt; &lt;span class="o"&gt;=&amp;gt;&lt;/span&gt; &lt;span class="s1"&gt;'COSINE'&lt;/span&gt;
&lt;span class="p"&gt;);&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h4&gt;
  
  
  Architecture Overview
&lt;/h4&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;┌──────────────────────────────────────────────────────────────────┐
│                    Legal Document Intelligence Platform          │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  ┌─────────────┐    ┌─────────────────────┐    ┌─────────────┐   │
│  │   Legal     │    │   Generative AI     │    │  Automated  │   │
│  │ Documents   │───▶│   ML.GENERATE_TEXT  │───▶│ Summaries   │   │
│  │ (Input)     │    │   AI.GENERATE_TABLE │    │ &amp;amp; Insights  │   │
│  └─────────────┘    │   AI.GENERATE_BOOL  │    └─────────────┘   │
│                     │   AI.FORECAST       │                      │
│  ┌─────────────┐    ┌─────────────────────┐    ┌─────────────┐   │
│  │   Legal     │    │   Vector Search     │    │  Semantic   │   │
│  │ Documents   │───▶│   ML.GENERATE_EMBED │───▶│ Search &amp;amp;    │   │
│  │ (Input)     │    │   VECTOR_SEARCH     │    │ Matching    │   │
│  └─────────────┘    │   ML.DISTANCE       │    └─────────────┘   │
│                     └─────────────────────┘                      │
│                                                                  │
│  ┌─────────────────────────────────────────────────────────────┐ │
│  │              Hybrid Intelligence Pipeline                   │ │
│  │         Combining Generative AI + Vector Search             │ │
│  └─────────────────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────────────┘
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h3&gt;
  
  
  Performance Results
&lt;/h3&gt;

&lt;p&gt;The implementation delivered impressive results across all metrics:&lt;/p&gt;

&lt;h4&gt;
  
  
  Processing Performance
&lt;/h4&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;ML.GENERATE_TEXT&lt;/strong&gt;: 6.99 seconds per document (3 documents processed in 20.98s)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;AI.GENERATE_TABLE&lt;/strong&gt;: 6.82 seconds per document (3 documents processed in 20.46s)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;AI.GENERATE_BOOL&lt;/strong&gt;: 0.48 seconds per document (3 documents processed in 1.44s)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;AI.FORECAST&lt;/strong&gt;: 1.29 seconds for 7 case outcome predictions&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;VECTOR_SEARCH&lt;/strong&gt;: 3.33-4.36 seconds per query (tested with 8 different legal queries)&lt;/li&gt;
&lt;/ul&gt;

&lt;h4&gt;
  
  
  Business Impact
&lt;/h4&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Document Summarization&lt;/strong&gt;: 99.2% efficiency improvement&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Manual Work&lt;/em&gt;: Lawyer reads 10-15 page legal document, identifies key issues, writes 3-sentence summary&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;BigQuery AI&lt;/em&gt;: 6.99 seconds for automated summarization&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Time Saved&lt;/em&gt;: 15 minutes → 6.99s per document&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Data Extraction&lt;/strong&gt;: 99.4% efficiency improvement&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Manual Work&lt;/em&gt;: Paralegal manually reviews document, identifies parties, extracts case numbers, court names, outcomes&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;BigQuery AI&lt;/em&gt;: 6.82 seconds for structured data extraction&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Time Saved&lt;/em&gt;: 20 minutes → 6.82s per document&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Urgency Detection&lt;/strong&gt;: 99.8% efficiency improvement&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Manual Work&lt;/em&gt;: Legal professional reviews document for deadlines, urgent matters, priority classification&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;BigQuery AI&lt;/em&gt;: 0.48 seconds for automated urgency detection&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Time Saved&lt;/em&gt;: 5 minutes → 0.48s per document&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Vector Search&lt;/strong&gt;: 540x faster than manual research&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Manual Work&lt;/em&gt;: Legal researcher manually searches through case law databases, reads multiple cases to find relevant precedents&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;BigQuery AI&lt;/em&gt;: 3.33 seconds for semantic similarity search&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Time Saved&lt;/em&gt;: 30 minutes → 3.33s per search&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Cumulative Impact&lt;/strong&gt;: 44.7 minutes saved for 3 document summarizations, 59.7 minutes saved for 3 data extractions, 14 hours saved for 7 case outcome predictions&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h4&gt;
  
  
  Quality Metrics
&lt;/h4&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Semantic Search Accuracy&lt;/strong&gt;: 55.1%-70.0% similarity matching for legal documents&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Meaning&lt;/em&gt;: When searching for similar cases, the AI finds relevant matches with moderate to strong semantic similarity&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Best Performance&lt;/em&gt;: 0.713 similarity score for "writ of mandamus" query (71.3% match confidence - strong similarity)&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Score Interpretation&lt;/em&gt;: 0.551-0.700 range indicates documents share related themes but are not identical&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Processing Reliability&lt;/strong&gt;: Consistent performance across all BigQuery AI functions&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Meaning&lt;/em&gt;: All 6 AI functions (ML.GENERATE_TEXT, AI.GENERATE_TABLE, etc.) delivered reliable results&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Document Volume&lt;/em&gt;: Successfully processed 1,000+ legal documents (tested with 3 document sample)&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Consistency&lt;/em&gt;: Stable performance across different document types and sizes&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Scalability Performance&lt;/strong&gt;: Handles large document volumes efficiently&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Resource Efficiency&lt;/em&gt;: Cloud-native architecture scales automatically&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Cost Effectiveness&lt;/em&gt;: Pay-per-query model with no infrastructure overhead&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Batch Processing&lt;/em&gt;: Optimized for handling multiple documents simultaneously&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  Real-World Legal Industry Challenges
&lt;/h3&gt;

&lt;h4&gt;
  
  
  Challenge 1: Managing High Volumes of Unstructured Data
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Problem&lt;/strong&gt;: Legal teams deal with vast amounts of unstructured data including emails, PDFs, chat logs, and multimedia files. This diversity complicates organization, retrieval, and analysis, leading to inefficiencies and increased risk of errors.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Solution&lt;/strong&gt;: Our BigQuery AI platform processes diverse unstructured legal documents through ML.GENERATE_TEXT and AI.GENERATE_TABLE, automatically organizing and extracting key information from complex legal texts.&lt;/p&gt;

&lt;h4&gt;
  
  
  Challenge 2: Ensuring Data Privacy and Compliance
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Problem&lt;/strong&gt;: Protecting Personally Identifiable Information (PII) within legal documents is paramount. Failure to properly redact sensitive information can lead to data breaches and non-compliance with regulations such as GDPR and HIPAA.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Solution&lt;/strong&gt;: The platform implements secure document processing with AI.GENERATE_BOOL for urgency detection and structured data extraction that maintains document integrity while identifying sensitive content.&lt;/p&gt;

&lt;h4&gt;
  
  
  Challenge 3: Handling Diverse Data Formats
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Problem&lt;/strong&gt;: Legal documents come in various formats, from structured contracts to unstructured case notes. Processing this diversity requires systems capable of handling multiple file types and extracting relevant information accurately.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Solution&lt;/strong&gt;: Our unified BigQuery AI approach handles diverse document formats through a single platform, using VECTOR_SEARCH for semantic understanding across different document types and ML.GENERATE_EMBEDDING for consistent processing.&lt;/p&gt;

&lt;h4&gt;
  
  
  Challenge 4: Overcoming Inefficient Document Management
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Problem&lt;/strong&gt;: Many law firms struggle with outdated document management systems leading to inefficiencies, multiple repositories, over-reliance on email, and unsearchable content resulting in lost files and missed deadlines.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Solution&lt;/strong&gt;: The platform provides centralized, searchable document processing with AI.FORECAST for predictive insights and comprehensive vector search capabilities that make all legal content discoverable and actionable.&lt;/p&gt;

&lt;h3&gt;
  
  
  Key Learnings
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;BigQuery AI Functions&lt;/strong&gt;: All 6 functions worked reliably with real legal documents&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Domain-Specific Prompts Matter&lt;/strong&gt;: Legal terminology and context significantly improved results&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hybrid Approach is Powerful&lt;/strong&gt;: Combining generative AI and vector search provides comprehensive coverage&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Performance is Excellent&lt;/strong&gt;: Sub-second processing for most functions with high accuracy&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Vector Search Effectiveness&lt;/strong&gt;: 0.551-0.700 similarity scores demonstrate moderate to strong semantic similarity&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scalable Processing&lt;/strong&gt;: Successfully handled 1,000+ documents with consistent performance&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Production Ready&lt;/strong&gt;: BigQuery AI functions are mature and reliable for enterprise use&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Business Value&lt;/strong&gt;: Real-world implementation proves the business value of AI in specialized domains&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Future Enhancements
&lt;/h3&gt;

&lt;p&gt;The platform has significant potential for expansion including diverse unstructured data processing, advanced clustering, multi-language support, API development, and industry specialization. The cloud-native architecture provides a foundation for scaling to international markets and developing custom legal domain models for enhanced accuracy and expertise.&lt;/p&gt;

&lt;h4&gt;
  
  
  Diverse Unstructured Data Processing
&lt;/h4&gt;

&lt;p&gt;&lt;strong&gt;Multi-Format Document Support&lt;/strong&gt;: Expand beyond case law to process diverse legal document types:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Contracts and Agreements&lt;/strong&gt;: Employment contracts, supply agreements, licensing deals, merger documents&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Legal Briefs and Motions&lt;/strong&gt;: Appellate briefs, motion filings, amicus briefs, trial briefs&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Regulatory Filings&lt;/strong&gt;: SEC filings, compliance documents, regulatory submissions&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Court Transcripts&lt;/strong&gt;: Oral arguments, deposition transcripts, hearing records&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scanned Documents&lt;/strong&gt;: Historical legal documents, handwritten notes, legacy case files&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Mixed-Format Archives&lt;/strong&gt;: Combining text, images, and structured data from legal proceedings&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Enhanced Data Pipeline&lt;/strong&gt;: Develop specialized processing workflows for each document type:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;PDF Processing&lt;/strong&gt;: Extract text from PDF contracts and legal documents&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Image OCR&lt;/strong&gt;: Convert scanned legal documents to searchable text&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Multi-Modal Analysis&lt;/strong&gt;: Process documents containing both text and visual elements&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Format-Specific Prompts&lt;/strong&gt;: Optimize AI prompts for different legal document types&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Cross-Format Similarity&lt;/strong&gt;: Find relationships between different types of legal documents&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;This expansion would demonstrate the full potential of BigQuery AI for processing truly diverse unstructured legal data, addressing the core challenge of handling mixed-format data that companies struggle to process effectively.&lt;/p&gt;

&lt;h3&gt;
  
  
  Conclusion
&lt;/h3&gt;

&lt;p&gt;Building this Legal Document Intelligence Platform with BigQuery AI demonstrates how AI can solve real-world business challenges in the legal industry. The combination of generative AI and vector search capabilities provides a powerful foundation for transforming how legal professionals work with documents.&lt;/p&gt;

&lt;p&gt;The results speak for themselves: &lt;strong&gt;99%+ efficiency improvements&lt;/strong&gt; and &lt;strong&gt;consistent performance&lt;/strong&gt; across all functions. This proves that BigQuery AI is ready for production use in specialized industry applications, delivering measurable business value and cost savings.&lt;/p&gt;

&lt;p&gt;Have you worked with BigQuery AI functions? I'd love to hear about your experiences and any questions you have about this implementation!&lt;/p&gt;




&lt;p&gt;&lt;em&gt;This project demonstrates the full potential of BigQuery's AI capabilities for solving real-world business challenges in specialized industry applications.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Tags&lt;/strong&gt;: &lt;code&gt;#bigquery&lt;/code&gt; &lt;code&gt;#ai&lt;/code&gt; &lt;code&gt;#legaltech&lt;/code&gt; &lt;code&gt;#machinelearning&lt;/code&gt; &lt;code&gt;#googlecloud&lt;/code&gt; &lt;code&gt;#vectorsearch&lt;/code&gt; &lt;code&gt;#generativeai&lt;/code&gt; &lt;code&gt;#businessautomation&lt;/code&gt;&lt;/p&gt;

</description>
      <category>bigquery</category>
      <category>ai</category>
      <category>vectorsearch</category>
      <category>legaltech</category>
    </item>
    <item>
      <title>🎉 ReUI v1.0.20: 9 New Premium Charts (100% Free!)</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Thu, 24 Jul 2025 13:58:34 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/reui-v1020-9-new-premium-charts-100-free-3cj7</link>
      <guid>https://dev.to/faizalsukaimi/reui-v1020-9-new-premium-charts-100-free-3cj7</guid>
      <description>&lt;p&gt;Hey developers! 👋&lt;/p&gt;

&lt;p&gt;I'm excited to share that we've just released &lt;strong&gt;ReUI v1.0.20&lt;/strong&gt; with a massive update: &lt;strong&gt;9 brand new chart components&lt;/strong&gt; that are completely free to use. No catch, no premium paywall - just high-quality, production-ready charts for your React applications.&lt;/p&gt;

&lt;h2&gt;
  
  
  What's New
&lt;/h2&gt;

&lt;h3&gt;
  
  
  📈 9 Line Charts
&lt;/h3&gt;

&lt;p&gt;We've added 9 modern line charts designed to present key metrics and insights. Each chart features unique layouts, data visualizations, and styling options perfect for dashboards, admin panels, and analytics pages.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Key Features:&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Sales overview with goals tracking&lt;/li&gt;
&lt;li&gt;Portfolio balance visualization&lt;/li&gt;
&lt;li&gt;Custom tooltips with percentage changes&lt;/li&gt;
&lt;li&gt;Gradient fills and modern styling&lt;/li&gt;
&lt;li&gt;Interactive dropdown menus&lt;/li&gt;
&lt;li&gt;Export functionality&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  📊 5 Area Charts
&lt;/h3&gt;

&lt;p&gt;Preview available with more chart variety for data visualization.&lt;/p&gt;

&lt;h2&gt;
  
  
  Technical Stack
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Built on shadcn/ui&lt;/strong&gt; - Seamless integration with existing projects&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Recharts powered&lt;/strong&gt; - Professional-grade charting library&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;TypeScript&lt;/strong&gt; - Full type safety and IntelliSense support&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Tailwind CSS&lt;/strong&gt; - Fully customizable with CSS variables&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;React&lt;/strong&gt; - Modern hooks and client-side rendering&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What Changed
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Migrated from OKLCH to Tailwind CSS variables&lt;/strong&gt; for better compatibility&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Updated Data Grid Component documentation&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;All charts are production-ready&lt;/strong&gt; with proper TypeScript types&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Dark/light mode support&lt;/strong&gt; out of the box&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  ⚡ Quick Start
&lt;/h2&gt;

&lt;p&gt;Install any chart with a single command:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;pnpm dlx shadcn@latest add https://reui.io/r/line-chart-1.json
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  🎯 Perfect For
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Admin panels&lt;/strong&gt; and dashboards&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Analytics applications&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Financial dashboards&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Data visualization projects&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Business intelligence tools&lt;/strong&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  🔗 Resources
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Live Preview:&lt;/strong&gt; &lt;a href="https://www.reui.io/blocks/charts/line-charts?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=v1.0.20-release&amp;amp;utm_content=line-charts" rel="noopener noreferrer"&gt;https://www.reui.io/blocks/charts/line-charts&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;GitHub:&lt;/strong&gt; &lt;a href="https://github.com/keenthemes/reui" rel="noopener noreferrer"&gt;ReUI Repository&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Documentation:&lt;/strong&gt; &lt;a href="https://www.reui.io/docs?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=v1.0.20-release&amp;amp;utm_content=docs" rel="noopener noreferrer"&gt;ReUI Docs&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  💭 What's Next?
&lt;/h2&gt;

&lt;p&gt;We're working on:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Bar charts and pie charts&lt;/li&gt;
&lt;li&gt;More interactive features&lt;/li&gt;
&lt;li&gt;Additional customization options&lt;/li&gt;
&lt;li&gt;Performance optimizations&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  🤝 Community
&lt;/h2&gt;

&lt;p&gt;We'd love to hear from you! Try out the charts and let us know:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;What you're building with them&lt;/li&gt;
&lt;li&gt;Any features you'd like to see&lt;/li&gt;
&lt;li&gt;Performance feedback&lt;/li&gt;
&lt;li&gt;Integration suggestions&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Drop a comment below or reach out on &lt;a href="https://twitter.com/reui_io" rel="noopener noreferrer"&gt;Twitter&lt;/a&gt;!&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Happy coding! 🚀&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;P.S. All charts are MIT licensed and completely free to use in both personal and commercial projects.&lt;/em&gt;&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Tags: #react #typescript #charts #ui #shadcn #recharts #tailwindcss #webdev&lt;/em&gt;&lt;/p&gt;

</description>
    </item>
    <item>
      <title>🎉 Metronic Angular v20 Boilerplate — 10 Demos Ready to Go!</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Tue, 15 Jul 2025 09:59:17 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/metronic-angular-v20-boilerplate-10-demos-ready-to-go-3mhk</link>
      <guid>https://dev.to/faizalsukaimi/metronic-angular-v20-boilerplate-10-demos-ready-to-go-3mhk</guid>
      <description>&lt;p&gt;If you’re building with Angular and want a head start on your next project, there’s some good news: the Metronic Angular v20 boilerplate is now available, and it comes packed with 10 ready-to-use demos.&lt;/p&gt;

&lt;h3&gt;
  
  
  What’s Inside?
&lt;/h3&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;10 complete demo layouts&lt;/strong&gt; — Explore a variety of dashboard and app designs, all styled with Tailwind CSS and Metronic’s clean UI.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Easy integration&lt;/strong&gt; — The boilerplate is set up for quick start, so you can focus on building features, not wrestling with setup.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Modern Angular&lt;/strong&gt; — Built on Angular v20, you get the latest framework improvements and best practices out of the box.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Metronic assets included&lt;/strong&gt; — All the styles, scripts, and design elements you need are already in place.&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  Get Started
&lt;/h3&gt;

&lt;p&gt;Ready to dive in? Check out the official integration guide and see all the demos in action:&lt;/p&gt;

&lt;p&gt;👉 &lt;a href="https://keenthemes.com/metronic/tailwind/docs/getting-started/integration/angular?utm_source=dev.to&amp;amp;utm_medium=social&amp;amp;utm_campaign=metronic&amp;amp;utm_content=angular"&gt;Metronic Angular Integration Docs&lt;/a&gt;&lt;/p&gt;




&lt;p&gt;Whether you’re prototyping, building a client project, or just exploring what’s possible with Angular and Metronic, this boilerplate gives you a solid foundation and plenty of inspiration.&lt;/p&gt;

&lt;p&gt;Happy coding!&lt;/p&gt;

&lt;h1&gt;
  
  
  Angular #Metronic #WebDevelopment #TailwindCSS
&lt;/h1&gt;

</description>
    </item>
    <item>
      <title>Metronic Angular v20 boilerplate is out—now with 10 demos ready to go! Get started: https://keenthemes.com/metronic/tailwind/docs/getting-started/integration/angular?utm_source=dev.to&amp;utm_medium=social&amp;utm_campaign=metronic&amp;utm_content=angular #Angular</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Tue, 15 Jul 2025 09:53:43 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/metronic-angular-v20-boilerplate-is-out-now-with-10-demos-ready-to-go-get-started-1pe6</link>
      <guid>https://dev.to/faizalsukaimi/metronic-angular-v20-boilerplate-is-out-now-with-10-demos-ready-to-go-get-started-1pe6</guid>
      <description>&lt;div class="crayons-card c-embed text-styles text-styles--secondary"&gt;
    &lt;div class="c-embed__content"&gt;
        &lt;div class="c-embed__cover"&gt;
          &lt;a href="https://keenthemes.com/metronic/tailwind/docs/getting-started/integration/angular?utm_source=dev.to&amp;amp;amp%3Butm_medium=social&amp;amp;amp%3Butm_campaign=metronic&amp;amp;amp%3Butm_content=angular" class="c-link align-middle" rel="noopener noreferrer"&gt;
            &lt;img alt="" src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fkeenthemes.com%2Fmetronic%2Ftailwind%2Fdocs%2Fdist%2Fassets%2Fmedia%2Fapp%2Fog-image.png" height="560" class="m-0" width="800"&gt;
          &lt;/a&gt;
        &lt;/div&gt;
      &lt;div class="c-embed__body"&gt;
        &lt;h2 class="fs-xl lh-tight"&gt;
          &lt;a href="https://keenthemes.com/metronic/tailwind/docs/getting-started/integration/angular?utm_source=dev.to&amp;amp;amp%3Butm_medium=social&amp;amp;amp%3Butm_campaign=metronic&amp;amp;amp%3Butm_content=angular" rel="noopener noreferrer" class="c-link"&gt;
            Metronic - Tailwind CSS Angular
          &lt;/a&gt;
        &lt;/h2&gt;
          &lt;p class="truncate-at-3"&gt;
            Integrate Angular with the Tailwind CSS based toolkit for dynamic web applications.
          &lt;/p&gt;
        &lt;div class="color-secondary fs-s flex items-center"&gt;
            &lt;img alt="favicon" class="c-embed__favicon m-0 mr-2 radius-0" src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fkeenthemes.com%2Fmetronic%2Ftailwind%2Fdocs%2Fdist%2Fassets%2Fmedia%2Fapp%2Ffavicon-32x32.png" width="32" height="32"&gt;
          keenthemes.com
        &lt;/div&gt;
      &lt;/div&gt;
    &lt;/div&gt;
&lt;/div&gt;


</description>
      <category>angular</category>
      <category>tailwindcss</category>
      <category>webdev</category>
      <category>announcement</category>
    </item>
    <item>
      <title>Keenthemes Metronic v9 (Tailwind version) with Symfony, Flask, and Django</title>
      <dc:creator>Faizal Sukaimi</dc:creator>
      <pubDate>Fri, 20 Jun 2025 07:07:51 +0000</pubDate>
      <link>https://dev.to/faizalsukaimi/keenthemes-metronic-v9-tailwind-version-with-symfony-flask-and-django-bi</link>
      <guid>https://dev.to/faizalsukaimi/keenthemes-metronic-v9-tailwind-version-with-symfony-flask-and-django-bi</guid>
      <description>&lt;p&gt;Hey everyone,&lt;/p&gt;

&lt;p&gt;I've written a few guides on integrating Keenthemes Metronic v9 (Tailwind version) with Symfony, Flask, and Django.&lt;/p&gt;

&lt;p&gt;Each one covers:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Full setup and asset handling&lt;/li&gt;
&lt;li&gt;Converting the Metronic HTML into Twig/Jinja templates&lt;/li&gt;
&lt;li&gt;Layout demos using both sidebar and header versions&lt;/li&gt;
&lt;li&gt;Real working code to get things moving quickly&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Check them out if you're working with any of these frameworks:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Symfony: &lt;a href="https://keenthemes.com/metronic/tailwind/docs/getting-started/integration/symfony?utm_source=dev.to&amp;amp;utm_medium=social&amp;amp;utm_campaign=metronic&amp;amp;utm_content=symfony"&gt;Symfony guide&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Flask: &lt;a href="https://keenthemes.com/metronic/tailwind/docs/getting-started/integration/flask?utm_source=dev.to&amp;amp;utm_medium=social&amp;amp;utm_campaign=metronic&amp;amp;utm_content=flask"&gt;Flask guide&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Django: &lt;a href="https://keenthemes.com/metronic/tailwind/docs/getting-started/integration/django?utm_source=dev.to&amp;amp;utm_medium=social&amp;amp;utm_campaign=metronic&amp;amp;utm_content=django"&gt;Django guide&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The full code is on GitHub here:&lt;br&gt;
&lt;a href="https://github.com/keenthemes/metronic-tailwind-html-integration" rel="noopener noreferrer"&gt;GitHub link&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Hope it helps. Let me know if you try it out or run into anything.&lt;/p&gt;

&lt;p&gt;#webdev #templates #symfony #flask #django #metronic #tutorial&lt;/p&gt;

</description>
    </item>
  </channel>
</rss>
