AI video automation
2025-2026
Client
Pega
AI-Powered Video Automation
Reducing personalized sales video delivery from 14 days to under 5 minutes
Role
Team
The Challenge
Pega’s sales and partner teams needed personalized video content that showcased the platform’s capabilities tailored to each client’s industry and use case. The existing process was entirely manual – from scriptwriting to screen recording to editing – and took an average of 7–14 days per video.
I was tasked with designing and automating this end-to-end workflow, leveraging AI to eliminate bottlenecks and dramatically reduce turnaround time.
Producing a single personalized sales video required coordination across multiple people and tools. A typical workflow looked like this: a sales rep would request a video, a copywriter would draft a script, someone would manually record branded screen captures of the platform, and an editor would assemble everything in After Effects. The result was high-quality but painfully slow – and it didn’t scale.
Iteration 1 - Semi-automated pipeline
Migration & Foundation
We started by auditing the existing application screens and identifying the most commonly used components. Rather than trying to build everything at once, we took an organic approach – creating components as they were needed in real projects, which ensured everything we built was immediately useful.
Intelligent Shot Matching
The system maintains a bank of pre-recorded video clips. When a script is generated, an LLM analyzes each section and selects the most relevant clips based on semantic matching.
Automated Editing via Plainly
The selected clips, script, and brand assets are imported into Plainly, a tool that renders finished videos from After Effects templates. This eliminated the manual editing bottleneck entirely.
Result: Average delivery time dropped from 7–14 days to under 48 hours.
The one remaining manual step: someone still needed to record short screen captures showing the platform branded for the specific client.
Iteration 2 - Fully automated
In the second iteration, we eliminated the last manual step by automating the screen recording process. The platform now generates client-branded screen captures programmatically, making the entire pipeline fully autonomous.
Result: A complete personalized video is now generated in under 5 minutes, end to end.





