
Decart
Decart is an AI company whose Lucy 2.0 model delivers real-time video transformation of live sports footage at under half a second latency; France Televisions named it as a vendor in its Roland-Garros 2026 RG Lab AI stack.
Last refreshed: 17 June 2026 · Appears in 1 active topic
Decart transforms live sports video in under half a second; which tournament is next?
Timeline for Decart
Supplied Lucy 2.0 model that transformed live sports footage at Roland-Garros in under half a second latency
Media's AI Pivot: France TV names its AI vendorsBackground
Decart is an AI company whose Lucy 2.0 model delivers real-time transformation of live sports footage at under half a second latency. France Televisions named Decart as a vendor in its RG Lab AI production stack at Roland-Garros 2026, disclosed on 11 June 2026, making it one of the first AI companies publicly credited for live generative video transformation in an active broadcast production context. The sub-500-millisecond latency is the threshold at which real-time broadcast use becomes technically viable, distinguishing Lucy 2.0 from post-production AI tools.
Decart operates in the generative video infrastructure segment, targeting broadcast and live-production workflows rather than consumer-facing content generation. Its positioning is at the latency-constrained end of the market: applications include live style transfer, visual enhancement, and on-air content transformation that must complete within the broadcast chain without introducing perceptible delay. France Televisions' Roland-Garros deployment is the company's most prominent public production credit to date.
The broadcast-vendor landscape for live generative AI is nascent. Decart sits alongside companies like Topaz Labs (remastering) and Gracia AI (volumetric capture) in what France Televisions' RG Lab disclosure reveals as a multi-vendor stack rather than a single-provider solution. The competitive set extends to established broadcast-technology companies integrating AI as a module rather than building AI-first from the outset.