We’re in the business of paying attention- to scenes, moments, & the things most systems skip past.
Roeve is a contextual advertising company built for Over-The-Top and Linear Broadcasting platforms. We work to make advertising fit the content it sits inside, using what is actually happening on the screen rather than anything about the person watching.
The industry spent twenty years asking who’s watching. We think that was always the wrong question.
Roeve started in 2024, when our founder was sitting through yet another ad break that made no sense — a loud, upbeat spot dropped straight into the middle of a funeral scene. It wasn’t a one-off. It happened constantly, on every channel, because no one in the chain — not the broadcaster, not the ad server, not the agency — was actually looking at the content itself. They were buying and selling slots based on who might be watching, never on what was actually on screen.
That was the whole insight, really. The industry had built extraordinarily sophisticated ways of profiling the viewer, and almost nothing to understand the moment. So we started building the thing that was missing — a system that reads a scene the way a person would, and uses that instead of a guess about who’s on the other side of the screen. What began as a frustration turned into the company we’re building today.
Ibhaan KudalkarFounder
Our Manifesto
For twenty years, the industry got better and better at knowing who was watching. More data, sharper profiles, deeper identity graphs. They all pointed at the same question: “Who is this person?” We started Roeve on a different question:
“What is this moment actually saying?”
It turned out to be the more useful one. An ad doesn’t work because it found the right person, it works because it arrived at the right moment. So we built a system that reads content instead of people: no viewer identity, no device graph, no behavioural history.
The harder problem was making that reading actually mean something. Content doesn’t arrive in neat categories- a scene in one language, one genre, one market carries an emotional grammar that a fixed taxonomy will always get wrong somewhere else. So instead of forcing content into boxes, we built something adaptive: a system that reads each scene on its own terms, and treats every rule — channel, territory, rights — as an input rather than an assumption baked in from day one.
None of that matters if no one can trust it. So every recommendation Roeve makes carries its reasoning with it- visible enough that a scheduler or an ad ops lead can see exactly why a moment was matched, and disagree with it if they should. We hold our own claims to the same standard: where a signal is strong, we act on it; where it’s weak, we say so.
And the call, always, stays human. Roeve recommends- it doesn’t place, traffic, or decide. A scheduler still builds the break. A compliance reviewer still clears what runs. We didn’t build this to remove that judgment. We built it to give that judgment better evidence, faster — and to make sure anything uncertain lands in front of a person, not a silent default.
Ready to see context in action?
We’re working with broadcasters and advertisers who are done guessing where an ad belongs. Let’s talk about what that could look like for you.