By Brian Thoman, CTO, WideOrbit

Artificial Intelligence (AI) has moved beyond the theoretical – it’s now an operational necessity. I recently participated in TVNewsCheck’s AI and Optimizing Advertising in 2025 webinar, where the panel explored how AI is already changing creative, planning, and the business rules that govern revenue forecasting and generation. The opportunity is enormous, but so are the risks if we don’t adapt fast.

Moderator Jon Accarrino framed the conversation early on, saying “[AI] is quickly becoming a vital tool for all areas of the media industry, especially advertising.”

Here’s what stood out to me during our conversation, and how we’re responding at WideOrbit.

Creative: Speed Without Losing the Story

Michael Vamosy (Defiant LA) captured the creative shift perfectly: “AI…helps you accelerate your ideas faster and further.” It’s an accelerant, not a substitute for taste or imagination, “A digital camera is not going to make you Ansel Adams with Photoshop.” Budgets are tighter, expectations are higher, and AI closes the gap between concept and proof.

Mary Rogers (Futuri) demonstrated that acceleration in action, using SpotOn inside Topline to show us auto-generated spec spots “from nothing more than just a business URL,” complete with scripts, voiceover, and music variants. As Mary put it, “The smartest broadcasting teams are…using AI creative tools to get in the door faster and win business they couldn’t have before.”

How WideOrbit is helping: Here at WideOrbit, we’re connecting those upstream creative and instruction systems directly into WideOrbit’s core trafficking environment, so when sellers or clients approve a concept, the correct creative and copy can flow into schedule assignment with fewer manual handoffs.

Planning Bias: “AI can’t see what it can’t measure”

Mary also walked us through Futuri’s sobering study. If, as she said, “Sixty percent of marketers adopt LLM-based planning, broadcast revenue could be cut in half within three years.” Futuri’s report showed that with LLM-based planning, “Linear TV averaged just 7% of the plan, versus 18% for CTV and OTT,” with some models excluding TV (and radio) entirely. Why? Because “AI can’t see what it can’t measure.” The publicly available, machine-readable evidence of broadcast advertising’s effectiveness is too thin relative to digital, resulting in AI bias in favor of more readily available digital data.

Caroline Giegerich (IAB) added market context, pointing out that, “Fifty-one percent [of advertisers] are currently using GenAI in digital video ad creation and 34% are planning to use.” GenAI-built or enhanced video is growing, from 22% (2024) to 30% (2025) to 39% (2026). Buyers want versioning, contextual relevance, and easier creative testing. Her recommendation for fairness in AI-assisted planning is to, “Make sure that broadcast data can be as standardized and machine readable [as digital] so that the planning tools can easily process it.”

How WideOrbit is helping:

• Sell-side planning agents. We’re building agentic workflows that use campaign goals and actual inventory availability to generate plan options sellers can act on quickly, closing the speed gap with digital.
• Attribution-aware prompts and data feeds. When broadcasters have outcome data, our planning agents can factor that into recommendations and produce machine-readable rationales. As results data accumulates, it can become a more impactful part of the equation.
• Open APIs. We’re publishing APIs in parallel with our AI features, so third-party planning, attribution, and creative systems can both contribute data and consume outcomes. As I said on the webinar, we believe in interoperability – we’re not holding the gate.

Operations: Fix the Bottlenecks, Then Automate

Mike Palmer (Sinclair) was blunt about pace, saying it’s “crazy…to think that a process that can take 24 to 48 hours to get content…to air is going to compete with digital.” Sinclair is applying AI where it pays back first: generating only the last three to five seconds for local tags across many versions; moving promos on demand instead of pre-building thousands; and – critically – using AI to turn unstructured emails and PDFs into structured traffic instructions. “We’re looking at AI to take those unstructured instructions and put them in a structured form… using fuzzy logic to bring those two things together.”

He also highlighted two industry-wide levers: standards (shared IDs and consistent processes) and measurement. His PSA to the group: “Support the NAB’s efforts…to sunset ATSC 1.0 and adopt ATSC 3 faster.”

How WideOrbit is helping:

• AI makegood agent. We’re building an agent that reviews viewership/estimate data, as well as what has been pulled from the log, and then offers makegood recommendations with choices that sellers can select for buyer review and approval. Makegoods won’t vanish, but the time and effort involved can be significantly reduced.
• Creative-to-traffic handoff. We’re wiring creative metadata and instructions from upstream tools into our core trafficking solutions, so assigning creative and standards and practices checks can be assisted by AI agents instead of email ping-pong.
• Agent-to-agent integrations. Yesterday, systems spoke via APIs only; tomorrow, systems will also talk through AI agents. We’re enabling both, so creative, planning, trafficking, and verification tools can coordinate without humans re-keying the same details multiple times, in multiple systems.
• APIs everywhere. Whether you use all WideOrbit systems or if ours are just a part of your stack, we’re making the APIs available so automation can work across your stack, not just within ours.

What to Do Next (Starting Now)

The webinar closed with practical advice worth amplifying:

• Educate your teams. Mike’s first step: Get everyone aligned on what AI is and isn’t. There’s a middle ground between “it will take every job” and “it can’t be trusted.”
• Start with measurable, boring problems. Normalize order emails and PDFs. Fix ID hygiene. Small wins compound.
• Feed the machines with proof. Mary’s challenge: “The stations that win will refuse to be invisible…They’ll flood the market with proof.” Publish case studies and results where LLMs can crawl them (websites, LinkedIn, PR wires) and tag them with industry terms so the data lands in the right knowledge graphs.
• Standardize. Caroline’s point: Make broadcast data as clean and readable as digital. That’s how we correct AI’s current planning bias.

AI is the Future – and the Future is Now

I used a Marvel analogy on the webinar for a reason: “Tony Stark built his first Iron Man suit by hand, then he taught Jarvis how to build them, and then he never built a suit by hand again.” We’re at that critical juncture with AI and advertising. Creative cycles are compressing from weeks to days to hours, and downstream systems must keep up. That’s why at WideOrbit, we’re focusing on automation where bottlenecks happen (makegoods, material instructions) and AI agents where speed and coordination matter (sell-side planning, inter-system workflows).

If we pair clean data and clear standards with practical AI, broadcasters won’t just keep up with digital, they’ll win on effectiveness and trust. And we’ll do it by working together, and without losing the human judgment that still decides what great advertising looks like.

Want to dive deeper into the discussion?

Watch the full webinar recording to learn more about how AI is transforming advertising in 2025 and beyond.

You can also contact us to learn more about WideOrbit AI Solutions.