Explore

The next generation of ads

Goal of the Document

There are many ways to improve ads. Here, we give 3 concrete examples through which we can solve these problems in new ways using generative AI.

Despite years of innovation, users hate ads. Why?

Ads aren’t relevant

Ads come at the worst possible time

Ads are too long; too frequent

Ads lack creativity & fail to capture my attention

Ads take too long to buffer, play - ruining my experience

Ads are a nuisance and just something to overcome; I don’t trust ads

We can make ads far more relevant.

We can in-paint elements of the product within the content so that the brand looks less out of place.

e.g. A retro-styled cap for a show set in the 1980’s. (example below)

Illustration 1: This promo from Stranger Things with a Nike product embedded into it. (Nike)

⁠

(model used: stable-diffusion-3)

We can reduce how jarring ads feel during transition.

We can make content & ads integrate deeper into each other.

For example, using AI, we can have the last frame of a video and an ad transition easily into each other.

Illustration 2: Keyframing from Luma and Stable Diffusion can reduce transition between videos.

⁠

(model used:

Luma Dream Machine⁠

)

We can craft ads with as much love as content.

The love / attention poured into content is seldom poured into ads.

But ads can look just as good as content:

Illustration 3: Let’s say that we are watching Planet Earth (a nature-oriented show). A simple brand logo in an app install campaign can be elevated into something as beautiful as Planet Earth.

⁠

(model used: Luma Labs + Keyframes)

Some other assorted ideas

Podcasts / Audiobooks → what if we can generate the ad in the voice of the podcaster / audio person?

This would make the ad far more trustworthy.

Can ads be as episodic as the content itself?

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.