Public TV

Live stories, turned into broadcast-ready clips.

Public TV's pipeline was not broken. It was built, professionally, for the 8pm bulletin, while the audience moved to feeds that never sleep. Their question: what can AI actually do here? Our answer: an autonomous newsroom, from live feeds to published clips, with one operator supervising.

operator supervising, not 5 to 8

~20 min

from break to broadcast

$1-2

per finished clip, self-hosted

10+

shorts a day

How it works.

We do the unglamorous work first. Then the AI work becomes obvious, scoped, and shippable.

The feed never stops. Production does.

A short-form clip touched editor, researcher, writer, producer, voice, video editor, motion designer, and reviewer in sequence. The stories aged faster than the pipeline could package them. That is not a talent problem. It is a shape problem.

Five to eight people in sequence, 45 to 120 minutes per clip
Most of the time spent on assembly, not journalism
Stories going stale while the package moves between desks

Editorial agents that find the beats.

The Hot Take loop: a Director agent discovers and curates stories near-instantly from live feeds. Parallel Researcher agents deep-dive the strongest candidates. A Writer turns research into script, a Stage Director adds shots and pacing, an Editor reviews, and a Sponsor agent reads the finished piece and places contextual ad segments at natural breaks. Mission Control shows an operator what every agent is doing.

Ingest, enrich, curate, script, voice, composite, review, package: one visible pipeline
Every stage adjustable and recoverable without restarting the job
Branded compositing for vertical and short-form output

The newsroom starts remembering beats.

A real newsroom does not think in individual headlines. Beats are tracked across days: durable, reusable story packages that know what happened before, what changed today, and what is still developing.

Related story events clustered as they arrive
Editorial runs and decisions tracked durably
Story packages reused across shorts, bulletins, deep dives, and sponsor inventory

Taste, quality control, human in the loop.

An operator cockpit where every stage can be inspected and re-rendered without restarting the job. Branded compositing so the output looks like it belongs to the channel, not to a template. Ultrarealistic talking heads at channel-grade quality, running on open-source models we self-host: that is what makes the economics work at daily volume.

Designed voices and lipsynced presenters, self-hosted on production servers
When the format demands cinema, the same brain drives Unreal Engine MetaHumans
Every stage observable, editable, and re-renderable

The human moves up the stack.

The human does not disappear; the human moves from assembly to judgment, sourcing, taste, and the publish decision. The newsroom catches beats as they break and publishes to Instagram, YouTube, and wherever the audience actually lives.

One operator supervising instead of five to eight in sequence
Shorts, bulletins, and long-form cut from the same story packages
The machine generalizes: sports, markets, regional-language news, branded explainers