ViralMoment, an AI intelligence layer over public TikTok, YouTube and Instagram video, was a hard sell to Gartner and Forrester because they saw it as social listening.
Elizabeth Arnold says more AI engineers can now build similar tools, which validates the category but makes it harder to hold.
Arnold says owning raw data will stop being a differentiator, so ViralMoment is building its own models for taxonomies, sentiment, emotion and visual recognition.
Creating a new category in video intelligence took years of persuasion
ViralMoment built a new category, AI analysis of public social video, and found that creating a category is slower and harder than competing in one. Text-based social listening misses what video carries, so the product needed analysts and buyers to see it as distinct. As AI makes similar tools easier to build, the company's lasting edge has to come from its own models and from layering video over text data.
Takeaways
Creating a category proved a harder sell than expected. ViralMoment was the first AI intelligence layer on public social video from TikTok, YouTube and Instagram. Analysts at Gartner and Forrester treated it as social listening, and it took about a year to reach the right people.
Text-only listening misses most of what video says. Arnold argues that gestures, faces, tone and music add meaning that text tools cannot capture. She cites a claim that 82% of internet content is video, which is the basis for the new category.
Buyers shifted from insight researchers to CMOs. The team first worked closely with insight researchers, who were moving into marketing organisations over about four years. It then repositioned the product toward CMO-level buyers rather than a roughly $100,000 sidebar tool.
Competition validates the category but erodes the edge. As AI spreads, more engineers can build similar tools. Arnold sees that as validation, yet expects raw data access to stop being a differentiator within about eighteen months.
Proprietary models and multimodal layering are the defense. ViralMoment is building its own AI models for taxonomies, sentiment and emotion, and logo and visual recognition. It also layers video over text sources such as Reddit to show how sentiment moves between platforms.
Partnerships and community beat LinkedIn for growth. Agencies, individual insight researchers and community groups have been the most sustainable channel, including sponsored events and dinners. Partners take time to get up to speed, but Arnold says the upfront payoff is dramatic.
A lean remote team leans on AI agents. At the Sprinklr acquisition the company had two full-time front-of-house staff and two engineers plus contractors. A Philippines contracting team covers hours while the US team sleeps, and internal agents include a weekly brand report builder.
Tap a name to go there · tap a line to see why they are linked