Headless Internet, Vibe Coding And Search’s Death: An AI Level Set For Broadcasters

The broadcasting industry is at a pivotal moment with artificial intelligence. Artificial intelligence is already reshaping how content is created, distributed and monetized. To explore how rapidly AI is impacting the local broadcasting industry, I spoke with leading AI consultant, Jon Accarrino, the founder of Ordo Digital and fellow TVNewsCheck columnist to explore how broadcasters are currently implementing AI while also discussing their roadmaps to future trends. Jon shared actionable insights for media organizations at any stage of their AI journey.  

Tom Sly: Any advice for broadcasters just starting their AI journey? 

Jon Accarrino: Jump in now. Don’t wait for perfection. Experiment relentlessly, learn as you go, and build a company culture that embraces innovation. But be wary of “free,” insecure AI tools. When a product, like ChatGPT is free, then you are the product, your data more specifially. So, try to deploy secure AI enterprise tools across the entire organization as soon as you can. Everyone from your interns to the CEO should have access to approved AI tools. Otherwise, your company will likely experience a data leakage problem.

Unfortunately, I’ve seen everything from sales budgets to HR data get fed into public AI tools. If you don’t give your teams the tools they need, then they turn to alternative resources.

TS: Are there examples of a costly mistakes that media companies should try to avoid? 

JA: Buying monthly seats for your employees with Copilot or ChatGPT is expensive, limits LLM variety, makes internal data access more challenging and lacks customization. Look for flexible enterprise solutions instead. If you need inspiration, both Graham Media and Morgan Murphy Media have publicly shared details on their enterprise deployments at multiple industry conferences.

Graham Media was able to reduce its monthly LLM costs by 90% without any notable impact to users by changing which models were connected to which apps. Limiting your organization to only use one solution, like Copilot, won’t give you that kind of flexibility.  

TS: When working with broadcasters to implement AI, is there a roadmap you follow?  

JA: It’s a phased approach. First, we typically conduct a workflow audit to identify opportunities. Next, we prioritize AI solutions that align with the broadcaster’s goals — like cutting costs, boosting engagement or unlocking new revenue streams. The key is to avoid “boiling the ocean.” Start small, measure impact and scale thoughtfully. 

TS: Media companies must have an AI policy. Is there an easy place to start creating your AI policy?  

JA: Absolutely. Begin with ethical guidelines how AI will be used responsibly in newsrooms, sales teams, HR, etc. Address bias mitigation, transparency in automated content and data privacy and security. Many organizations use frameworks as a starting point. But keep it loose and flexible. Your policy should evolve alongside your AI adoption. And don’t be scared to tear it up and start over. You don’t want your AI policy to cripple your growth and innovation.  

TS: What AI apps do you see bringing the greatest value to broadcasters?  

JA: Versioning apps are the most requested category with our news-based media clients. Once a media org creates a story, they can use AI to help create multiple versions of that content. Convert a broadcast script to a web article, push alerts, promos, social media content, etc.

Combined with article metadata generation and SEO, these are a huge time saver for busy newsrooms and create opportunities for human staff to focus on more important tasks. If you haven’t already, look at NOTA. An analysis of its suite of AI tools for newsrooms showed a 52% average reduction in time spent to publish article content.

TS: AI is already great and generating text, what about images and video content?  

JA: It’s incredible to see how far AI-powered image and video production has advanced in just a few years. Visual creative tools like Flux, Runway, Veo 3 and Midjourney can generate amazing, lifelike imagery.

From weather explainers to station image spots, a variety of visual AI creative tools are already being used by innovative broadcast groups. AI can assist, but human oversight is critical for accuracy, trust and connection.

As Michael Vamosy from Defiant LA explained, your project needs to be “grounded in reality” or you’ll lose the audience. Highly creative tasks like visual storytelling still require a focused human touch.  

TS: For media companies already using AI, what are some bigger projects that they should consider?  

JA: Once you have your AI tools in place, then start with the annoying things in your workflow, like tasks that are repetitive and time-consuming. Automated AI transcription, closed captioning, data analysis and metadata tagging are low-risk, high-reward entry points. 

For example, for one of my clients, I’m building multiple tools to help their investigative teams. Deploying agents to automatically scrape social media posts from politicians or analyze and organize messy reports from government websites, can be huge timesavers for busy newsrooms.    

AI-powered advertising creative is another great opportunity. Check out companies like Streamr.ai and Creatopy to help your sales teams.

TS: Can AI’s costs be offset by incremental revenue?  

JA: Yes, but it requires strategic investment. The ROI becomes clear when AI is tied to measurable business outcomes, like expanded hyperlocal coverage, audience growth or reduced production times. There are also revenue-supported AI solutions, like Dappier’s agentic-powered AskAI experiences or Stringr’s AI-automated VideoGen animations. These tools can be deployed with sponsorships or ad units, making them immediate revenue generating solutions for publishers.

But if you need immediate cash to fund your AI innovations, then consider licensing your video archives to AI companies for training purposes. Video, that’s where their focus is right now. Companies like OpenAI and Google have already mastered text and audio. Video is the next frontier for AI. I’ve been able to help clients sell their video content for up to $6 per minute. Think about how many hours of video content that you have in your archives, and then do the math.

TS: What exceptional AI implementations have you seen in broadcasting?  

JA: At this year’s Local Media Consortium conference, Lindsay Stewart, Stringr’s CEO, debuted its new PUBGEN tool that can use AI to create video clips from a written news article. The tool uses images from the article and/or licensed news photo libraries, writes a script, creates an AI voice over and generates a video version of the article. And it only takes a few moments to create. Just imagine being able to create a video for every article on your website with just a few clicks. It’s incredible. Solutions like this can help enable magazine, newspaper and radio companies to instantly become video publishers.

Broadcasters should also check out what meteorologist Amy Freeze is doing. She created a very realistic AI avatar of herself and now delivers custom weather updates to clients around the world. She is leveraging her brand, her IP, and making it scalable in ways that were never possible before. In the time it takes a typical TV meteorologist to create and deliver a single weather report on camera, Amy can deliver dozens of videos.

TS: What’s on the horizon for AI in broadcasting?  

JA: It’s a little scary. We are on the edge of AI being able to power headless experiences across the internet. All AI needs is a data source and user input. How audiences consume content almost becomes secondary. Allow me to explain:

• NotebookLM can create a custom podcast episode about any topic that you want.
• Meta is developing an AI-powered ad platform where advertisers don’t need creative or targeting data anymore, just a bank account. AI does the rest.
• For the first time ever, Google Search on the iPhone is declining. User habits are shifting to prefer voice assistants and chatbots over Googling to find answers. 
• At NAB, Futuri shared survey results showing that 45% of local TV audiences think that an AI algorithm would be better at selecting and curating news stories than a human. 
• Vibe coding has already shown us AI can create on-demand user interfaces, apps and websites. Will AI eventually get to the point where it can create apps and websites on-demand? Is this the end of apps and websites? It may not be long until the user’s “vibe” could determine the UI/UX on the fly. [Editor’s Note: Accarrino has an article for TVNewsCheck about this topic being published soon.]
• And the level of AI-powered personalization and automation that will be possible for audiences has barely even started. During my panel at Programming Everywhere, Stringr Founder and CEO Lindsay Stewart said, “Your audience doesn’t want the news, they want their news.” 

Let all of this sink in for a minute. 

What happens when AI is better at both curating content and building sustainable delivery interfaces than humans? 

The most valuable assets that local media companies have, are data and personality. Protect and monetize them carefully. 

TS: Thank you, Jon, this has been informative and hopefully will inspire those that have not started their AI journey or will help provide guidance to those that are well on their way.