Brian Casel - Videos
Back to Channel
Building a Custom AI News Agent with RSS and Telegram
Brian shows how an AI agent named "Vale" scans RSS feeds of key industry players, filters them for relevance, and delivers a curated Markdown report to Telegram every morning. π **Your Builder Bri...
Multitasking With Agents: My 2026 Workflow
Almost a year ago, I made a video on multitasking with Claude Code. I knew it was possible back then β but it didn't stick, for me or most of us. It's only now, in 2026, that I've actually become c...
Can you trust OpenClaw with your business?
An honest comparison contrasting the polished reliability of the Claude ecosystem against the scrappy, chaotic, but breakthrough potential of OpenClaw. π **Your Builder Briefing (free)** https://b...
Build Your Next Startup Entirely From Your Phone
A high-speed workflow showing how to use a GitHub template on a mobile browser to spin up a new repo and immediately start coding it with Claude. π **Your Builder Briefing (free)** https://builder...
Where Claude Design actually fits
Most Claude Design tutorials show you the controls and some flashy demos. This video is different β I put it through my real workflow as a builder to find where it actually fits. I'll show you the ...
Why Claude isn't ready for fully autonomous agents?
A critical breakdown of why Claude's ecosystem currently functions better as a collaborator rather than a fully autonomous team member compared to OpenClaw. π **Your Builder Briefing (free)** http...
Creating a Consistent Visual Style with AI Agents
Brian demonstrates a sophisticated workflow where Claude acts as a creative director to develop concepts, which are then passed to the Gemini API for high-fidelity, on-brand image generation. π **...
AI Writes My Docs By Reading My Commits
Brennan demonstrates how he uses a tool called FernDesk to automatically generate customer-facing help documentation by analyzing his specific code files and git commits. π **Your Builder Briefing...
We build fast. But does it work?
We're shipping code faster than ever, but the bottleneck has shifted to QA. Unit tests and integration tests are a solid first layer, but there's still a gap between "the tests pass in CI" and "a r...
How Claude's New Auto Memory Actually Works?
A breakdown of the difference between managing your own static instruction files versus letting Claude's new Auto Memory silently learn from your course-corrections over time. π **Your Builder Bri...
Hire When You Have a Job, Not a Task β The Rule That Changes Everything
Brian applies a real-world hiring principle β only hire for recurring roles, not one-off tasks β to AI agents, and explains why the low cost of agents dramatically lowers the threshold to start del...
4 Agent Skills I Use for Marketing
I walk through four custom agent Skills I've built to handle the marketing side of my business. Each one started as a repeatable process I was doing by hand β now they're delegated to my agents. Th...
How to Plan Features in Claude Code
Brian demonstrates the "Shape Spec" workflow, showing how to use plan mode to have the AI interview the user about a new feature (a search bar) to create a robust plan before implementation begins....
A visual approach to spec-driven development
I've been impressed by this tool for more extensive planning, especially with it's built-in visual tools and plan to task tracking. Use Nymbalyst for free - The visual workspace for building with ...
The Skill AI Can't Replace
AI can build anything now. The harder question is what deserves to be built. I break down why restraint is the most important skill in AI-first development, then give you a concrete framework for p...
Claude Code on Mobile: The Complete Guide
You can build from anywhere with Claude Code β your couch, a coffee shop, on a train. But most people have only seen one piece of the mobile story. In this video, I walk through every scenario for ...
How to build AI agent processes that actually scale?
A strategic breakdown of why wrapping repeatable business processes into portable Markdown "skills" protects your workflows from AI platform lock-in. π **Your Builder Briefing (free)** https://bui...
Claude Code changed recently
Claude Code in 2026 is not what it was a year ago. In this video, I walk through three features β auto-plan, auto-memory, and voice mode β that might look like small quality-of-life improvements on...
Connect Claude to Your Business Data
A specific tactic for connecting Claude to event stores and analytics, allowing non-technical leaders to query business data (like completion rates) directly through Slack. π **Your Builder Briefi...
OpenClaw vs. Claude for Running an Agent Team
I've been running my business across two AI ecosystems β OpenClaw for autonomous agent delegation and Claude's family of products for deep collaboration and creative work. In this video I break dow...
Stop Fixing Small CSS Tweaks Yourself
Arvid and Brian discuss why they ask Claude to make even tiny color changes rather than editing the code manually, arguing that the AI understands dependencies and side effects that a human might m...
Stop Making Mockups Start Making PRs
Brian explains why static mockups and specs are obsolete compared to generating "Claude Code prototypes" directly in Slack that run on the actual codebase. π **Your Builder Briefing (free)** https...
The JSON First Design Workflow
Brennan shares his new design process: asking the AI to generate realistic JSON sample data first (to account for edge cases), and then generating the UI mockups based on that data. π **Your Build...
How to create JOBS for OpenClaw agents
The number one question after my last OpenClaw video was: what are you actually having your agents do? Instead of just listing tasks, this video gives you a framework for hiring agents like real em...
Let the AI Interview You Before You Code
A look at how the AI now natively asks clarifying questions to shape the specβa feature that previously required custom frameworksβand why this "spec-driven" approach is critical. π **Your Builde...
My Multi-Agent Team with OpenClaw
I ordered a Mac Mini just to run OpenClaw β and set up a team of 4 AI agents to help me run my business. In this video, I walk through every decision I made: hardware, security, costs, chat interfa...
I Built an AI Tool to Turn Voice Notes into LinkedIn Posts
Colleen demonstrates a tool she is building live that monitors a local folder for new voice notes and uses Claude to instantly generate formatted LinkedIn posts based on the audio transcript. π **...
Why You Lose Control When AI Writes Your Code
Arvid explains "Comprehension Debt"βthe risk that the developer's mental model of the software drifts away from the actual codebase when AI writes code without the developer understanding the "theo...
Claude Code in Slack changes how teams SHIP
I came across an article about how Anthropic uses Claude Code in Slack internally β and it changed how I think about the ROI of adopting AI with your team. Your engineers are shipping faster with C...
Build your own marketing tools with Claude Code
Colleen Schnettler is a Ruby on Rails developer who worked at ClickFunnels, runs SimpleFileUpload (https://www.simplefileupload.com), and is now launching a technical marketing consulting business....
Why I Stopped Using AI Frameworks in 2026
Brian argues that complex AI frameworks are now "overkill" and explains why he has reverted to using native, vanilla tools for 90% of his daily work. π **Your Builder Briefing (free)** https://bui...
The Easiest Way to Train Your AI on Your Codebase
Arvid demonstrates his workflow of recording a video walkthrough of his app as if training a new employee, then having Claude convert the transcript into a structured technical documentation file (...
Brennan Dunnβs Team of (AI) Developers
In this 2nd installment of 'Builder Stories', Brennan Dunn, founder of RightMessage (https://rightmessage.com), a website personalization tool, shows us his sophisticated system of custom AI agents...
Agent OS v3: Leaner & Smarter for Building in 2026
Agent OS v3 is here β rebuilt from the ground up for how we actually build in 2026. I stripped out 70% of the framework. What's left: a leaner system for discovering, documenting, and injecting yo...
Arvid Kahlβs SaaS is 98% Coded by Claude
Trying something new on the channel! This is episode #1 of a new series called 'Builder Stories' where I'm inviting fellow builders to show us how they build and tell the story of their transition...
Claude Code is all you need in 2026
Claude Code in 2026 is not what it was when it launched. After a year of building frameworks, custom setups, and elaborate systems to make AI coding work, we've come to a point now where... most of...
Replacing my n8n workflow with a Claude Code Skill
I spent a week building a complex n8n workflow for AI image generation β then scrapped it and rebuilt the whole thing as a Claude Code skill in 30 minutes. Here's why automation without intelligenc...
This one AI skill multiplies all the others
Knowing when to use Claude vs ChatGPT, terminal vs IDE, subagents vs main contextβthat's workflow orchestration. Build this intuition and you're not just reacting to AI tools, you're directing them...
Design OS: The Missing Design Process for AI-First Development
AI coding tools build fastβbut the results are a hodgepodge. Disconnected screens. Generic UIs. No cohesion. And re-prompting just makes it worse. Design OS fixes this. It's a methodical design pr...
Cursor's New Visual Editor: When It's Useful (And When It's Not)
Cursor's new visual editor lets you tweak CSS visually and apply changes directly to code. I break down when it's actually useful and when you'll want to reach for something else. π **Your Builder...
Stop Prompting in Circles (Do This Instead)
The secret to great AI-generated code isn't better promptsβit's better specs. Here's how to let AI interview YOU with clarifying questions so your agents build exactly what you envisioned, first tr...
Build with paralell agents in Cursor 2.0
Watch me use Cursor 2.0 to build a feature with multiple agents building in parellel, then compare their work and choose a winner. π **Your Builder Briefing (free)** https://buildermethods.com - Y...
I want to like Google Antigravity, but...
I was genuinely excited about Google Antigravity's fresh ideas for the agentic IDEβmulti-project management, multi-agent orchestration, and that clever screen recording review process. But when I a...
Taking Cursor's new Composer 1 model for a spin
Watch me build using Cursor's new 'Composer 1' model that dropped with Cursor 2.0. This thing is fast! π **Your Builder Briefing (free)** https://buildermethods.com - Your free, 5-minute read to ...
Claude Opus 4.5 vs. Google Gemini 3: Design & build an app
Let's compare Claude Opus 4.5 and Google Gemini 3 to see how they stack up when it comes to planning, designing frontend and backend for a new app! Plus how I evaluate new frontier models when the...
A short history of devs adopting AI (it happened fast)
rom tab-completion to autonomous agentsβhere's the 3-year timeline that changed everything about how we build software. If you've been skeptical about AI coding, you were actually right to be... un...
Breaking a Spec down into Tasks using Agent OS v2
How I break a feature spec down into a task lists using Agent OS v2. π **Your Builder Briefing (free)** https://buildermethods.com - Your free, 5-minute read to keep up with the latest tools & wor...
Spec a new feature using Agent OS
How I shape and plan a new spec for a new feature using Agent OS v2. π **Your Builder Briefing (free)** https://buildermethods.com - Your free, 5-minute read to keep up with the latest tools & wor...
From AI skeptic to UNFAIR advantage
Your skepticism about AI coding was justified. But the shift that's happened over the last 12-24 months changed everythingβand your experience is now your unfair advantage. I break down 5 skills t...
How Developers Should Use Claude Code Skills
After a week of testing Claude Code Skills, I've found two clear patterns where they actually make sense. First, they excel at utility tasks that pop up unpredictably - like commit messages, code r...