Case study · AdTech · client: a startup linked to Record TV

Media planning system

Given a budget and a campaign goal, the system assembles a radio and TV media plan. An OpenAI model composes the matrix of possible insertions; the user reshapes the plan on a dashboard where every value is recalculated as it changes; and the approved plan is sent straight to execution. Built by the NetDev team for a startup linked to Record TV.

Status
Delivered · no longer in commercial operation
Client
Startup linked to Record TV
My Role
Product Owner
Stack
Firebase, OpenAI API
Core Focus
LLM-composed plans, live recalculation, approval flow

The challenge

A startup linked to Record TV wanted an app that showed advertisers the best placement options for a campaign within the budget they set.

Key engineering decisions

A plan from a budget and a goal

The user states budget and campaign goal; the system sweeps the possible radio and TV insertions and crosses the best results within the brief, with an OpenAI model composing the matrix of possibilities.

A dashboard that recalculates as you build

The plan is edited on a dashboard where every change triggers the recalculation of values, so the planner sees the cost of each decision while making it.

Approval and hand-off to execution

The finished plan goes through approval inside the system and is sent to execution from there, without being re-typed into another tool.

Results

1 dayto close a media plan — it used to take a week of human work
minutesfor the system to assemble the plan from the brief

What’s left for people is adjustment and validation.

Stack

FirebaseOpenAI API