How AI Let Me Do the Work of 10 People for Pennies
Five Real Projects That Transformed How I Run My Media Business
The conversation around artificial intelligence in radio and the music industry usually swings between hype and fear. On the ground, it’s a lot more practical. For operators, programmers, publishers, and business owners, AI works best as a quiet assistant in the background handling repetitive work, accelerating analysis, and allowing one person to do what used to take an entire team.
This past week was a good example of that. I used AI across several parts of my business including archives, transcription, backups, competitive research, and domain management. Each project solved a real operational problem and eliminated hours of manual work.
Here’s what that actually looked like.
Sorting Through Nearly a Million Archived Files
Over the years I’ve accumulated a massive archive of documents. Interviews, drafts, transcripts, and notes stored across multiple drives. At this point the collection is approaching a million files.
Many of the older documents were created long before I had consistent naming systems in place, which makes it difficult to identify what’s inside a file without opening it manually.
Instead of doing that, I wrote a Python script that used AI to analyze the structure of the text inside those files. The script looked for question-and-answer patterns to determine whether a document was likely an interview.
Once those interviews were identified, I ran another pass to extract the names mentioned inside the documents so I could quickly see who the conversations were with.
From there I created a final step that pulled metadata from the files themselves to determine the actual creation date. This was important because files that move across drives often lose reliable modification timestamps. Pulling the original metadata helped place the interviews back into the correct timeline.
In a short period of time I was able to locate interview material across a massive archive that would have taken days or even weeks to sort manually.
Transcribing Video Interviews From My Local Terminal
Another project involved transcribing video interviews and other recordings.
Instead of relying on subscription transcription platforms, I installed Whisper locally and ran the transcription process directly from my terminal. Once the transcripts were generated, I ran them through ChatGPT to clean up punctuation and formatting.
One key instruction was making sure ChatGPT did not change the essence of the content. If you don’t give that direction, language models will often rewrite sentences or alter phrasing while trying to improve readability.
By instructing it to only fix punctuation and formatting, the transcripts remained faithful to the original spoken content while becoming far easier to read and work with.
The result was accurate transcripts produced at a fraction of the cost of most transcription services.
Building a Rotating Backup System in the Cloud
Reliable backups are critical, but many backup solutions become expensive if they are not managed carefully. Storage costs can climb quickly when backups are tied to servers or traditional hosting systems.
This week I built a rotating backup system in the cloud that solves that problem.
The system maintains three recent copies of my data at all times. Each time a new backup is created, the oldest one is automatically removed. That means there are always three recovery points available without storage growing endlessly.
Instead of relying on traditional server-based backup systems, I discovered a cloud storage approach that allows this entire process to run for pennies a month.
Once configured, the system runs automatically and stays completely separate from my primary infrastructure.
Comparing My Site Against Another Site
I also used AI to compare one of my websites against another site operating in the same space.
The goal was straightforward. I wanted to see where my site might be missing coverage and where it was already performing well.
Instead of manually reviewing hundreds of pages, AI analyzed the structure of both sites, looking at content categories, coverage depth, and other structural differences.
The results provided a clear overview of strengths and gaps. Some areas showed opportunities for expansion, while others confirmed that my site already had strong coverage.
That kind of analysis normally takes a significant amount of time. With AI assisting the process, it can be done quickly enough to actually influence strategy.
Evaluating the Value of Domains I’ve Been Holding
Like many entrepreneurs who have been operating online for years, I’ve held onto a number of domain names from older projects and ideas.
This week I used AI tools to analyze those domains and determine whether they still had value. The process looked at domain history, backlink patterns, and authority signals.
One thing that became clear is that domains left unused often accumulate spam backlinks over time. However, domains that previously hosted legitimate content may still retain meaningful authority.
Another reality is that holding domains long term has its own economics. Registration costs increase almost every year, and domain marketplaces often place advertising on parked domain pages to generate revenue for themselves.
Many of those platforms will also encourage you to list the domain for sale so they can act as the middleman between the buyer and the seller. If someone shows interest, they charge a fee and say they will contact the owner.
In my experience, I’ve rarely been contacted through those systems. After seeing how that process works over the years, I now prefer to sell my own domains directly instead of relying on a marketplace acting as the intermediary.
Where This Is Going
What stood out this week is how quickly operational problems can be solved once the right AI workflows are in place.
Tasks that once required multiple people such as sorting archives, transcribing interviews, managing backups, analyzing competitors, or evaluating digital assets can now be handled by one operator with the right systems.
Most media businesses, radio companies, and digital publishers are sitting on the same kinds of inefficiencies. Large archives that can’t be searched. Video content that hasn’t been transcribed. Domains that aren’t being used strategically. Backup systems that cost too much or aren’t reliable enough.
AI isn’t replacing the operator. It’s amplifying the operator.
When those systems are built correctly, one person can move at a speed that used to require an entire team.
Kevin Ross is the founder and publisher of Radio Facts, covering the business, infrastructure, and strategy behind radio, music, and digital media.
If you were born between 1930-1979