I Wasted $40K on an AI Agency. Here's What I Built Instead.

The episode discusses how non-technical founders can leverage AI, focusing on the journey of Aaron Anderson, an eight-year veteran of the link-building agency Linkpitch.io. Anderson's story highlights the transformative power of AI, especially for those without a technical background, and offers insights into navigating the evolving landscape of AI implementation in business. Anderson, a non-technical founder, initially spent weeks failing to automate a simple team reporting task using various no-code tools. However, when he turned to Claude Code and asked it to build the solution in Python, it was completed within an hour and hosted on GitHub Actions, working perfectly. This experience was a "Claude pill" moment for him, realizing the potential of AI when communicating in its "native language" of code, rather than through other tools. He now views AI as a "superpower" that allows him to build solutions for almost any problem. His journey into AI was partly driven by a recent downturn in his business. After consistent year-over-year growth, 2024 saw a revenue decrease. This prompted him to re-evaluate his business operations, particularly the challenges of training new team members, which could take a year and carried uncertainty about AI's future impact on SEO. He decided to shift towards becoming an "AI-native business," where operations are built on AI to automate much of the work, reducing reliance on extensive human training. Initially, Anderson was a slow adopter of AI, viewing ChatGPT and similar technologies as overhyped. However, once committed to becoming AI-native, he sought external help. He hired an agency that promised to rebuild all his systems to be AI-first, quoting $30,000. Despite initial hopes, the project went past its timeline and over budget, with the agency requesting more money. During this period, Anderson began experimenting with Claude Code independently. As his own AI building skills improved, he started to question the agency's value proposition, which centered on building software cheaply using AI. He realized he might be better off building solutions himself. This led him to rebuild existing systems in code to gain confidence. This experience taught him the importance of understanding AI and its capabilities before engaging external agencies. He realized that while AI could build anything, the "architecture" of the solution was crucial and often required experienced human input. He likened it to building a house: the construction doesn't take the most time, but the architectural design does. He also learned about the need for "finetuning" prompts when using AI within systems to achieve desired results. Anderson later found a highly recommended developer to work with, acknowledging his own weaknesses in architectural design. He emphasizes that while AI can get you far, especially in building, the strategic and architectural aspects still benefit greatly from human experience. Regarding the concept of an "AI-native" business, it means fundamentally integrating AI into daily operations. For Anderson, this involves automating repetitive tasks that were previously done manually. For instance, he created a tool using Playwright that automatically logs into Ahrefs and Semrush, downloads weekly reports, and compares them against outreach efforts to identify built links—a process previously done manually. The discussion also touched on the broader implications of AI for businesses, particularly in marketing. The host, Dan, shared his own experience of becoming AI-native by integrating AI into his podcast operations. He described a "hive" where all business intelligence, from listener preferences to past episode data, is accessible and queryable by AI, informing decisions on content positioning and audience engagement. This makes data analysis more accessible, leading to data-driven decisions that might not have been pursued before due to cost or effort. Regarding team adoption, Anderson hasn't mandated AI use but has encouraged it. He believes that repetitive tasks currently done by humans will eventually be handled by AI, allowing humans to be repurposed for more strategic roles. He has one team member experimenting with Claude Code to implement front-end changes. Looking ahead, Anderson draws a parallel between the current AI revolution and the early days of self-driving cars, suggesting that while AI will bring significant changes, the "Armageddon" scenarios might not unfold as quickly as some predict. He believes foundational SEO principles, like the importance of mentions and links, will remain relevant, even as people increasingly use LLMs for search. His focus is not on predicting the future but on leveraging current AI capabilities to position his business for whatever comes next. For founders looking to get started with AI, Anderson advises embracing the "tinkering" phase. He compares it to the early days of self-hosting WordPress, where non-technical individuals learned complex technical skills due to the promise of building their own websites. Now, AI acts as a mentor, guiding users step-by-step through the process of building solutions. He shared an anecdote where he, a non-developer, used Claude to push code to a server in 30 minutes, a task that would have taken an experienced developer an hour. This illustrates how AI significantly accelerates the learning curve and execution for non-technical individuals. The key takeaway is that the promise of AI, much like the early internet, is about enabling better businesses by allowing founders to build and automate solutions that were previously inaccessible due to cost or technical barriers.