Multimodality

The multimodal model works not only with text, but with several “modalities” at once: image, video, sound. It can both understand (describe what’s on the screen) and generate (make a video based on the script).

For marketing, this has broken down the boundaries between specialties. Previously: a copywriter writes, a designer draws, a videographer films, a sound engineer does voice-over - four people. Now one person manages everything through a multimodal stack: text, image (Nano Banana Pro), video (Veo), voice (ElevenLabs).

I measured the practical effect on my projects: content production fell by about 40% without loss of quality. Not because “AI did everything,” but because task transfers between people disappeared.