Build With AI

I sit down with Leon van Zyl, who ran a web design company for 10 years and now teaches over 700 people how to build real applications with coding agents. We walk through his exact workflow for building professional, client-ready websites using Google Stitch for the design system and Claude Code for the build — no coding skills required. Leon shows the difference between a one-shot AI-generated site and what you get when you front-load the design system before touching code. By the end, you'll have a repeatable workflow for going from design concept to finished website — including custom AI images that match your brand.

Links Mentioned:
Google Stitch: https://stitch.withgoogle.com
Claude Code: https://docs.anthropic.com/en/docs/claude-code
Cursor: https://www.cursor.com
Next.js: https://nextjs.org
Stitch MCP Server Docs: https://stitch.withgoogle.com/docs/mcp
Stitch Skills: https://stitch.withgoogle.com/docs/skills

Timestamps
00:00 – Intro
00:06 – What you'll learn: design systems for client-ready websites
02:04 – Jumping into the screen share
02:25 – The problem: one-shot AI websites look terrible
03:52 – The Stitch workflow result: side-by-side comparison
07:32 – Starting from a vanilla Next.js project
08:50 – What is Google Stitch and how to get started
10:00 – Prompting Stitch with brand details, fonts, and colors
13:00 – Why design systems matter for coding agents
16:00 – Iterating on the homepage before building more pages
17:44 – Sharing Stitch designs with clients for approval
21:23 – Setting up the Stitch MCP server in Claude Code
23:18 – What an MCP server actually is (simple explanation)
25:56 – Pulling the design system into your project
28:47 – Memory files: Claude.md vs Agents.md explained
33:24 – Converting the Stitch design into a working website
35:06 – Installing the Stitch React Components skill
41:11 – When to use this workflow: client work vs personal projects
44:27 – Viewing the finished website vs the Stitch mockup
48:27 – Downloading and converting images to WebP for performance
53:46 – Generating custom AI images with Nano Banana Pro
58:14 – Final result with branded AI-generated hero image
01:00:48 – Key takeaways and wrap-up

FIND ME ON SOCIAL
X/Twitter: https://x.com/coreyganim
Instagram: https://www.instagram.com/coreyganim/
LinkedIn: https://www.linkedin.com/in/coreyganim/

FIND LEON ON SOCIAL
YouTube: https://www.youtube.com/@leonvanzyl

Claude Code just dropped routines, and when I first saw the three options — routine, schedule, and loop — I had no idea when to use which. In this video, I break down all three modes in plain language and then demo building a routine from scratch using an X trend scanner I already had in my skill library. You'll see the full setup: GitHub connection, system prompt config, trigger options, and connector permissions. By the end, you'll know exactly which mode to use for any automation you want to build and how to get your first routine live.

Links Mentioned:
Claude Code Routines: https://claude.ai/code/routines
GitHub: https://github.com

Timestamps
00:00 – Intro: The confusion with Claude Code's new features
00:20 – The three modes: routine, schedule, and loop
00:50 – When to use each mode (simple breakdown)
01:10 – Using Claude Code to find a skill to demo
01:45 – Selecting the X scan routine
02:00 – GitHub requirement for routines explained
02:18 – Creating a new routine at claude.ai/code/routines
02:35 – Naming your routine and connecting your GitHub repo
02:50 – Dropping in your system prompt
03:05 – Setting the schedule trigger (10am Eastern daily)
03:20 – Configuring connectors and permission sets
03:40 – Running on demand and reviewing the output
04:10 – Full recap of all three modes

Key Points

* Claude Code has three automation modes — routine (runs on Anthropic's cloud, set it and forget it), schedule (needs local access, also available in Cowork), and loop (immediate execution in the CLI). Picking the wrong one means your automation won't work the way you expect.

* Routines require a GitHub repository. You can't create one without it, so get your repo connected before you start.

* The whole setup happens at claude.ai/code/routines — no coding required beyond having a system prompt in markdown format ready to paste in.

* Routines support three trigger types: time-based schedule, GitHub event, or API call. That gives you flexibility depending on whether you want it cron-based, push-based, or fired from another system.

* The X trend scanner ran automatically and surfaced several trending topic candidates with suggested content angles — zero manual research, just results ready to use.

FIND ME ON SOCIAL
X/Twitter: https://x.com/GanimCorey
Instagram: https://www.instagram.com/coreyganim/
LinkedIn: https://www.linkedin.com/in/coreyganim/

FIND NICK ON SOCIAL
X: https://x.com/NickSpisak_
LinkedIn: https://www.linkedin.com/in/nicholasspisak/
YouTube: https://www.youtube.com/@UCusMlMDWZS0uOlhgL0M_q-A

Grab the free Google Doc to build and deploy your first Claude Managed Agent in 60 seconds: https://return-my-time.kit.com/2872b904f5

I sit down with Nick Spisak, an AI agent builder who deployed his first Claude Managed Agent the same week Anthropic released the capability, to break down exactly what this platform is, who it's for, and when it makes financial sense to use it. We walk through the four user personas Claude Managed Agents is built for.

Links Mentioned:
Corey's X article: https://x.com/coreyganim/status/2042286607449874527
Nick's X article: https://x.com/NickSpisak_/status/2041949191887262164

Timestamps
00:00 – Intro
00:05 – What is Claude Managed Agents?
00:29 – Architecture: decoupling tools, sessions, and orchestration
01:15 – Who managed agents is and isn't for
01:33 – The four user personas breakdown
03:58 – Why AI Tinkerers should stick with Claude Code subscription
04:21 – Free Google Doc: build your first agent in 60 seconds
05:28 – Platform-as-a-service model explained
05:43 – Cost comparison: Managed Agents vs Agent SDK
06:09 – Live look inside platform.claude.com
06:23 – Workbench, sessions, and output walkthrough
08:16 – Sessions: tool calls vs. finished output explained
08:31 – Analytics dashboard and real cost breakdown
09:36 – When the ROI math works (and when it doesn't)
11:18 – Next steps: articles on X, how to get started
12:04 – Outro

Join the Build With AI community - weekly AI implementations, live coaching, and templates built for non-technical entrepreneurs: https://www.skool.com/buildwithai/about

FIND ME ON SOCIAL
X/Twitter: https://x.com/coreyganim
Instagram: https://www.instagram.com/coreyganim/
LinkedIn: https://www.linkedin.com/in/coreyganim/
YouTube: https://www.youtube.com/@coreyganim

FIND NICK ON SOCIAL
X: https://x.com/NickSpisak_
LinkedIn: https://www.linkedin.com/in/nicholasspisak/
YouTube: https://www.youtube.com/@nickspisak_

Download Nick's free second brain skill -- one-click install that sets up your entire Obsidian vault and ingestion workflow inside Claude Code: https://return-my-time.kit.com/286e11f7e6

I brought on Nick Spisak to build a complete second brain live -- start to finish, in under 20 minutes. By the end of this episode, you'll know exactly how to build your own second brain -- and you can grab Nick's free skill to do it in one click.

Timestamps
00:00 - Intro and Karpathy's viral second brain tweet
00:24 - What the second brain concept is
03:28 - Obsidian Web Clipper: scraping pages into your vault
04:41 - Nick's free skill: wizard, ingest, query, and lint commands
05:53 - Live demo: running the setup wizard in Claude Code
08:38 - How many vaults to manage: personal vs. business
10:30 - Opening the vault and exploring the file structure
12:21 - Graph view: seeing connections between your data
15:38 - Ingest command: raw data into organized wiki
16:29 - Automating ingestion on a cron schedule
19:12 - Compounding value and syncing the vault across devices
20:19 - Pruning the vault with the lint command
21:53 - Your data set as a moat in the AI age

Key Points

* Karpathy released a framework for building an LLM knowledge base. Nick turned it into a free Claude skill with a guided setup wizard, ingest, query, and lint commands -- works across Claude Code, Codex, Gemini CLI, and more. One-click install, no coding required.
* The system runs on three tiers: raw (brain dump), wiki (AI-organized knowledge base), and outputs (answers from querying). Drop files into raw, run ingest, and the AI maps everything into structured wiki entries with relationship graphs.
* You can automate ingestion using Claude Code's loop feature so the vault stays current without manual work. Pair with Obsidian's paid sync tier and a note captured on your phone is indexed before you're back at your desk.
* The lint command health-checks your wiki for outdated entries and missing connections -- and tells you exactly what to clip next to close the gaps. The wiki tells you what it doesn't know yet.
* Day zero this thing is basic. Day 90 it's a company asset no competitor can copy. Your private knowledge base is the foundation for every agent and skill you build -- and nobody else will have it.

Join the Build With AI community - weekly AI implementations, live coaching, and templates built for non-technical entrepreneurs: https://www.skool.com/buildwithai/about

FIND ME ON SOCIAL

X/Twitter: https://x.com/coreyganim
Instagram: https://www.instagram.com/coreyganim/
LinkedIn: https://www.linkedin.com/in/coreyganim/
YouTube: https://www.youtube.com/@coreyganim

FIND NICK ON SOCIAL

X: https://x.com/NickSpisak_
LinkedIn: https://www.linkedin.com/in/nicholasspisak/
YouTube: https://www.youtube.com/@nickspisak_

Eliot Prince is an SEO specialist with 10+ years of experience ranking businesses at the top of Google. He's now deep in the AI world, coaching businesses on how to do SEO faster and better using AI tools. In this episode, Eliot walks through how he's using Perplexity Computer to build repeatable SEO machines — systems that audit websites, generate deliverables, and even push changes live to WordPress, all from a single conversation.

We run a live audit on ReturnMyTime.com, watch three audit modules execute in parallel, and dig into the results — competitive analysis, keyword gaps, technical fixes, and auto-generated page copy. Eliot also demos a LinkedIn lead machine he built that reads DMs, scores leads, and drafts replies automatically using Perplexity's Linkup API connector.

The big takeaway: Perplexity Computer runs 19 LLM agents in the cloud simultaneously, meaning you can build systems that would traditionally take weeks of manual work or expensive SEO tool subscriptions - and you don't need to write a single line of code to do it.

Find Eliot YouTube's channel here:@princeeliot
Join the Build With AI community here: https://www.skool.com/buildwithai/about