← All case studies
Sri MandirSri Mandir
Devotion & culture · AppsForBharat

From 2–3 hours per video to close to 100 a day.

How Sri Mandir's content team replaced hand-built, scene-by-scene AI video with a SurgeGrowth pipeline — script in, finished videos delivered to Slack, ready to ship to Meta.

~5×
Videos per editor, per day (1–2 → 7–8)
~100
Videos shipped per day
< $2
Per 30-sec video, on Veo 3 Lite
The challenge

Every video was a manual, scene-by-scene grind.

Sri Mandir produces a steady stream of devotional content for Meta — talking-head recitations, explainers and short-form clips. But making each video was slow, hands-on work. An editor wrote a prompt for every scene by hand, generated one scene at a time, and manually fixed whatever came back wrong.

The in-house workflow was fragile, and it kept breaking. Worse, there was no clean way to regenerate a single scene and re-merge it into the finished cut — one small fix meant rebuilding the whole video. Start to finish, a single video took an editor 2–3 hours. On a good day, that was 1–2 videos per person.

What we built

One pipeline, from script to Slack.

We moved the entire process onto SurgeGrowth. The editor's job shrank to the two steps that actually need human judgment — the brief at the start and the final edit at the end. Everything in between runs itself.

Submit a script and a first frame
The editor enters the script and uploads the first frame of the character. That's the entire input — no per-scene prompt writing.
The workflow triggers
Generation runs in the background, scene by scene, on its own. The editor moves on to other work instead of babysitting one clip at a time.
The finished video lands in Slack
When the run completes, the video is posted straight to the team's Slack channel. No prompting hassle, no manual generation.
Regenerate any scene in one click
Not happy with a scene? Regenerate just that scene and re-merge the video in a single click — no starting over.
Quick edit, then ship to Meta
The editor makes quick final edits and ships the video to Meta. Runs are often triggered overnight — by the time the team logs in the next morning, the videos are already done.
The impact

More output, at a fraction of the cost.

The change was a step-change, not an increment. Per-editor output went from 1–2 videos a day to 7–8. Across the team, Sri Mandir now produces close to 100 videos a day — much of it generated overnight and waiting when the team arrives.

Cost fell just as hard: a 30-second video now runs under $2 on Veo 3 Lite. That economics is what makes scaling on Meta viable — the team can flood the channel with on-brand devotional content instead of rationing it.

Apps we built for Sri Mandir
AI Talking Head Realistic Character Images Automated video editing Create variations

Want this for your team?

Tell us what you're producing by hand today. We'll scope a build.

Book a call →