Episode 667 AI and Agency Life With Ryan Waterbury
Show Highlights
In this episode of the Agency Chat segment, host Rob Cairns and co-host Ryan Waterbury dive deep into the practical realities, challenges, and security concerns surrounding the integration of Artificial Intelligence (AI) in modern digital agencies. Moving past the “hype,” they explore how AI impacts cybersecurity patching, software development, content quality, and agency cost structures.
Thank You To Our Sponsor
Thanks you to our sponsor, All-in-One WP Migration by ServMask. Export your entire WordPress site in one click, import it anywhere. No server access needed, no command line, works with every hosting provider. Free on wordpress.org, Pro extension at servmask.com.
Show Notes
Key Takeaways
1. The Security Surge and the AI Link
- The Patch Tuesday Influx: June 2026 saw Microsoft issue over 300 security patches, with additional emergency patches required for actively disclosed vulnerabilities.
- The Linux Backlog: Linux creator Linus Torvalds has noted that the bug reporting queue has become nearly unmanageable.
- The Core Cause: Both hosts argue that AI is driving this surge. While AI is actively discovering legacy bugs that have existed for over a decade (such as recent cPanel vulnerabilities), it is also creating a massive influx of duplicate bug reports and introducing new vulnerabilities through AI-generated code.
2. The Danger of “Vibe Coding”
- Low-Quality Code Output: Tools like Replit and Lovable frequently generate code with underlying security flaws.
- The Human Element: Non-developers or average creators are using simple prompts to generate software packages and selling them directly to consumers, resulting in products filled with security risks and bugs.
- Best Practices: Ryan emphasizes treating AI as a “junior developer” or assistant. True security relies on adhering to frameworks (like Laravel or WordPress plugin standards) that have baked-in security guardrails, rather than deploying unchecked raw AI code.
3. Client Use and Agency Operations
- Good Implementations: Using built-in AI agents for automated internal workflows (e.g., updating CRM data fields when a lead reaches a specific milestone) remains an excellent use case.
- The Content Problem: Clients frequently submit raw AI-generated blog posts that lack brand voice and actual value, which can ultimately harm SEO performance rather than help it.
- The Rising Cost of AI: The belief that AI will continually save agencies money is a misconception. The cost per token is rising, platform limits change daily, and utilizing proper API integrations requires a significant financial investment.
4. Spam, Scams, and Local AI Models
- AI-Driven Fraud: The sophistication of phishing schemes, API key theft, and grandparent phone scams has escalated heavily due to AI-generated voices and scaling capabilities.
- The Case for Local Models: To counter rising cloud costs and secure sensitive client data, Ryan runs open-source models (such as Chinese models like DeepSeek or Minimax) locally on his Mac hardware. This maintains privacy without sacrificing context or capability.
Memorable Quotes
“We are definitely in the camp that AI has made life and business easier, but it has also made parts of it significantly harder—especially when managing security across a massive pile of websites.” — Rob Cairns
“Enter a prompt that says: ‘Make ChatGPT, make no mistakes, enter.’ It writes something, puts it out there, and people sell it. But guess what? It’s full of bugs and security problems.” — Ryan Waterbury
PocketCasts
Radio Public
Spotify
Stitcher
Apple Podcasts
Amazon Music
