Some ad campaigns hit the mark. Others not so much.
But the planning, reporting and analysis needed to build, assess and optimize TV ad campaigns based on what’s working and what’s not, historically can mean a manual and time-consuming process. And even with effort, it’s not always easy to quickly or clearly gauge what elements of a TV ad made viewers lean in or tune out.
But alongside investments in machine learning and AI compute power, iSpot is looking to inject AI agents to take out some of the leg work for brands and turn its trove of ads, audience and outcomes measurement data – alongside persistent consumer ad surveys – into actionable insights for video ad creative through a new conversational agentic AI platform dubbed SAGE.
Launched this week, SAGE was developed over the past two years with active input and testing from major brands including General Motors and Airbnb, for a total of 12 customer trials in private beta.
The first wave of product availability focuses on ad creative, with two main AI agents at launch: The Creative Insights Analyst for diagnosing ad performance and the Creative Planning Assistant that generates strategic recommendations with both historical campaign and real-time insights.
Later this year SAGE will introduce AI agents for competitive intelligence, audience, performance and optimization.
Underpinned by iSpot’s datasets and surveys, a goal is to speed up workflows by enabling teams to quickly zero in on and understand the key creative elements in TV ads that have (or have not) resonated with viewers and drive metrics like purchase intent. Those insights then help inform, guide and accelerate the campaign development processes, where the platform can also provide recommendations for new ideas. AI-generated insights are also meant to provide more assurance that ad creative efforts will achieve the consumer connection and help drive the KPIs brands desire (aka, be effective).
SAGE employs AI agents and conversational two-way communication via prompts in a ChatGPT-style for video advertising, but iSpot CEO Sean Muller told StreamTV Insider that it isn’t just a generalist AI platform or Large Language Model (LLM).
AI and agentic AI are all the buzz right now, but in order to be effective for brands and advertisers, platforms and AI models needs to be grounded in accurate data and trust.
And while SAGE does use OpenAI and available LLMs in the background, when it comes to things like measurement, the ads themselves and recommendations into what is or isn’t working, Muller said the platform is tightly trained to only trust proprietary iSpot data – of which there is plenty.
iSpot’s already a software vendor to advertisers that use it to measure, optimize and plan creatives and audience, with an aim to help deliver on the business result they want to achieve.
SAGE is the next evolution of that, per Muller, tying together the vendor’s existing datasets and AI tech to make easier the arduous tasks of creative and behavioral data pulls from earlier campaigns and analysis to know what is or isn’t working.
“You really just ask the question, and it will go do all the work and provide back the data and the measurement and the analytics and also the recommendation,” Muller explained. “Really what we've done is we've built our own large language model that really focuses in on our data, on our trusted measurement.”
Specifically, SAGE taps into iSpot’s big data from every frame of advertising campaigns, alongside persistent survey-based consumer testing and audience and outcome measurement across the broader ecosystem. iSpot has nearly 3 million ads in its catalog with full transcriptions of each and covers 130,000 products, 80,000 advertisers and 216 million metadata points. On the creative assessment data side, ad testing includes over 500 survey responders per ad with a set of standardized questions.
iSpot will add its audience and outcome data to SAGE later this year.
Through those measurement datasets and surveys, the vendor aims to offer brands a holistic view of the TV ad marketplace, spanning 185 TV networks, 500 publishers and data from tens of thousands of distinct brands.
The persistent creative survey testing feeds into the AI platform to shed light on consumer perceptions and give brands better understanding into the “why” behind creatives that hit or miss the mark.
SAGE boasts AI-powered frame-by-frame analysis and access to ad-level ACE Scores (likability, attention, watchability) and purchase intent metrics that are tied directly to specific creative themes within the ads.
Here’s a sample image of SAGE that shows an example of a prompt brands could enter and get immediate feedback – in this case, which Super Bowl 2026 ad was the most likeable and why did consumers like it?
Advertisers can then get back information about specific elements that audiences liked or didn’t, such as themes, celebrities and so on.
Using the frame-by-frame analysis, together with consumer surveys, SAGE categorizes ads into themes and super-themes, and then scores themes against metrics such as likability, attention and purchase intent.
SAGE also supports advanced audience segmenting, with the ability to filter results by specific demographics and high-value “intender” (or those consumers planning to buy) groups.
In a demo for StreamTV Insider, the platform zeroed in a specific ad spot from last year’s Super Bowl as the most likable and provided insights into why. Responses included information about what consumers liked, such as: the ad was inspiring and uplifting, emotional connection, high quality cinematography, it successfully tackled a serious health topic while maintaining an uplifting tone and so on.
It can also pull overarching or more granular themes and filter by industry.
Speeding processes with trusted data
In terms of ad creative data, SAGE extracts metadata and storylines from every frame across video ads (2.5 million ads) and compiles themes from every verbatim survey response (approximately 100 million verbatims).
All of that requires considerable computing power, for which iSpot said it made significant investment in NVIDIA AI servers.
And if computers need a lot of resources to extract and compile the massive datasets, one might image the amount of manpower such tasks would require.
According to Muller, generally speaking, it takes 8-12 weeks to develop new concepts, test them and run analytics, whereas with SAGE that timeline shortens to 1-2 weeks.
As for speeding processes, the platform is both pulling data faster and providing analysis that would’ve been done manually before, while also offering the ability to plan for new ads based on what works.
“This will actually write new ideation. It will write scripts and allow brands to develop creative much faster that’s grounded on actual metrics of what’s worked for them in the past and even what’s worked for other advertisers in the space,” Muller said.
He also emphasized how critical that trust in the underlying iSpot data informing the SAGE-delivered insights is to advertisers, noting the platform is also built with guardrails and privacy protections in place.
General Motors, a pilot tester and partner in developing SAGE, gave endorsements to both notions of reducing time to improve creatives and the need for trusted datasets in AI models used for TV ad investments.
“The market has made it clear that a trusted model, grounded in expert data, is the only AI they want for their video investments,” said Miles Drayton, global director of Marketing Intelligence at General Motors. “iSpot SAGE delivers on this by pairing the productivity gains of AI with a partner who knows their data better than anyone, eliminating the doubt and manual busy work that has plagued campaign optimization for years.”
And notably, by extracting metadata and aspects like storylines, characters, settings and product details from TV ads, the agentic AI platform can link specific creative elements to campaign performance – helping brands have more assurance in where to direct creative efforts to effectively connect with consumers as they seek to achieve goals.
“The primary goal for advertising is driving better connections with our customers and conveying information. iSpot SAGE is a great example of how we’re investing in innovations that drive better emotional resonance and recall,” said Kyler Blackmore, Media Director at Balance of Nature, in another client endorsement for the platform.
In the pipe, iSpot intends for SAGE to help with production workflows, as well as competitive intelligence with multi-brand analysis to run in-depth comparisons against competitor ads and themes.
The workflow automation function is still in beta, but iSpot said users will be able to instantly generate data-driven creative briefs, script outlines and storyboard directions to speed up the time from insight to production.
“iSpot brings a massive repository of what people feel about ads, what creative elements and narratives were used and where those ads were seen. The development of SAGE promises to expand our potential to forge new connections, using actual intelligence, which means it's data that we can trust,” said General Motors’ Drayton.