AI engines (like ChatGPT, Claude or Gemini) are redefining the way we search for information. Video, although a dominant and omnipresent format on the web, remains today a blind spot for these new response engines. As a result, millions of hours of content with high added value (demonstrations, interviews, reports) are excluded from this new research economy.
As part of the 2026 edition of Viva Tech, Aive announced the global launch of a new tool which allows videos to be played by LLMs.
An unprecedented technological alliance with NVIDIA
Aive relies on a technological combination: its proprietary MGT (Multimodal Generative Technology) technology combined with NVIDIA Nemotron open source models. This synergy makes it possible to analyze more than 25 simultaneous signals in a video (scenes, speakers, emotions, spatial context) and to precisely link what is said orally with what is displayed on the screen.
“Our technology doesn’t just do basic description. It tracks the elements in a spatio-temporal manner: we analyze the actions, their sequence, their exact duration. Moreover, this understanding ability works just as well on completely silent videos. HAS”, explains Rudy Lellouche, Co-founder and CPTO of Aive, met at Viva Tech.
To make a video understandable for an LLM, Aive will generate corresponding text.
“We act as real translators for AI. Our role is to break down the videos into multiple themes and generate the corresponding text, in order to make this visual content intelligible and “digestible” for LLMs. he continues.
Why AIs will never play videos on their own
One might wonder why the AI giants do not directly integrate this “viewing” capability. For Rudy Lellouche, the answer is above all economic and legal.
“Multimodal analysis of a video is an extremely cumbersome process and today, it would cost far too much for LLMs. Our role is to begin their work by translating these videos into multiple textual themes to make them “edible”. In addition, this resolves the issue of copyright, since we provide them with verified content whose exploitation is authorized by brands and media.” he explains.
All these extracted elements are then structured in a Video Content Graph, an intelligent mesh which interconnects all the subjects addressed by a brand or media over time, thus creating a database directly usable by conversational agents.
Extend the lifespan of content
Until now, the video distribution strategy was mainly limited to social networks, where the lifespan of content is extremely short. With Generative Engine Optimization (GEO), Aive offers brands and media the opportunity to free themselves from this time constraint.
“Today, the lifespan of a video is very limited by the ephemeral nature of social networks. Our discoverability solution allows us to overcome this classic consumption curve. We offer the possibility of monetizing and promoting this content in the long term, even when the initial audience peak has passed. HAS”underlines Rudy Lellouche.
The structuring of this data also transforms old cost centers, namely video storage, into profit centers. For the media engaged in a standoff with the tech giants, this is an opportunity, according to the co-founder:
“The media today are trying to charge LLMs for the use of their text articles, but it’s complicated to charge for something that is already easily accessible on the web. On the other hand, video archives are by definition inaccessible to these AIs. By making these videos readable, the media hold a real card to play in negotiating and monetizing access to this well of knowledge. HAS”
This is also a crucial issue for content creators. Today’s biggest influencers limit their presence to YouTube, TikTok or Instagram, but have absolutely no media existence in text form. Aive’s technology could allow them to exist outside of these platforms and finally be recommended by response engines.
Reconciling humans and algorithms
By creating content that feeds AI engines, we can fear the arrival of a two-faced Internet: a version intended for humans and a version dedicated to robots. However, Aive’s approach is focused on the end user. The objective is not to hide invisible texts on ghost sites, but to create real media coverage around the video, by generating articles that benefit both classic SEO and LLMs.
“We must not design content only for algorithms, but also for humans, while ensuring that they are perfectly understood by AI. This is the very essence of our solution: maximizing visibility for everyone. HAS” specifies the co-founder.
As Google gradually transforms into an AI engine, the challenge for companies is no longer just to be well ranked on a results page, but to be cited directly by artificial intelligence as a source of authority. As Rudy Lellouche sums it up perfectly: “In the age of generative engines, not being discovered by these AIs is simply not existing. HAS”
