CBS AI PRODUCTION PIPELINE

CBS wanted to scale social content without scaling production resources. An early AI tool could identify and rank potential clips from a full episode.

It didn't know what made a clip good.
The early tool could generate clips, but it lacked the editorial judgment, brand knowledge and workflow logic required to turn them into publishable CBS content.
That gap created another problem: the creative teams didn't trust it.

Experienced producers make hundreds of tiny editorial decisions without thinking about them. I worked directly with engineers as the de facto Production and Editorial Lead to identify those decisions and translate them into product requirements the system could understand.
To most people, this clip looked ready to publish. I saw three seconds we couldn't afford: a one-second exterior shot followed by nearly two seconds before the character speaks.
For short-form, the clip needed to begin with the character, not the scene.

I translated those editorial decisions into rules the system could follow:

Scene Boundaries: Fade-to-black and audio shifts helped identify exactly where scenes should begin and end.
Tone Mapping: Music cues and BPM helped distinguish serious moments from content appropriate for upbeat treatments.
Asset Guardrails: CBS logo placement established safe zones for automated text and graphics.
THE DIFFERENCE:
The system learned to identify the moment that mattered and begin the clip at the exact second the content started.

Once clipping worked, I expanded the workflow to everything required to publish: captions, copy, graphics, QC, and scheduling.
Eventually, a full episode could become scheduled social content through one workflow connected directly to Sprinklr.

Testing with users from first-time editors to veteran producers exposed the unwritten rules social teams took for granted.
Instead of teaching everyone those rules, we built them into the product, from platform safe zones to brand and formatting requirements.

I personally trained 50+ CBS marketing employees, from social managers through executive leadership.
Once teams saw the tool could handle the repetitive work without creating more cleanup, resistance shifted to adoption—freeing producers to focus on the creative work they actually wanted to make.

CREATIVE JUDGMENT, BUILT TO SCALE.
What started as a clipping tool became a workflow that embedded editorial judgment, brand standards, and social production rules directly into the product.
The technology didn't replace creative judgment. We built creative judgment into the technology.

RESUME