Empowered 6M+ video editors worldwide to create faster and gain traction with great enjoyment
Turn selected clips from your past videos into a new draft with AI
Users can select clips from their existing video library, and OpusClip’s AI will automatically generate a cohesive video draft, saving time on manual editing and making it easy to repurpose past content into fresh videos.
Product Type
New feature, AI,
2C and SMB
Duration
Dec 2024 - Mar 2024
(4 month)
Team
JJ Li (Head of Product)
Responsibility
Research, Prototyping,
User Testing
With over 6 million active users, Opus Clip is known for turning long videos into short clips. Now, it's expanding its capabilities to help creators streamline more of their workflow, improve productivity, and create more value from their existing content.
Jake - YouTuber
Jake runs a finance channel covering stock analysis. He often reuses past footage for updates, comparisons, or recaps, and needs a fast way to turn archived videos into fresh content.
Maya – Digital Marketer
Maya creates social content for a startup, repurposing webinars and long videos into short, punchy promos. She looks for tools that save time and keep content on-brand.
Ryan – Podcaster
Ryan runs a podcast with over 100 episodes. He frequently assembles themed compilations or highlight reels and needs an efficient way to pull clips from his video library.
Users are now able to select clips from their library, have AI generate a structured narrative, and receive a draft video ready for refinement. This workflow accelerates the editing process while still giving creators control over the final outcome.
Research
Finding the problem
Early Prototypes
To get specific feedback and validate solution early
Snapshots of early prototypes on output layout variations, to validate early and iterate based on evidence
Before we assumed that users would start with writing prompt and let AI find relevant clips and put them together based on the prompt. However, this workflow presents users with a lot of friction and is not intuitive for our users
Workflow Before
User Flow
Design Decision - Searching
Exploring Cart Layout
Option 1: Horizontal tray below search bar
Option 2: Vertical cart on the right
Option 3: Horizontal tray at the bottom
✅ Horizontal tray at the bottom
7/8 users
like the bottom tray because it aligns with video editor’s mental model of assembling the clips horizontally
Design Decision - Automation vs. Manual
Two options
I explored two options that varies in level of automation
Option 1: Manual flow — users adjust AI-selected clips before AI proceeds to generate the draft
❌ Require too much manual effort and time consuming
5/8 users
think manual adjust clips is time-consuming, and wanted more automation in storyline regeneration.
❌ Confusing workflow and interface
7/8 users
felt confused about the workflow and the elements on this screen and required explanations
Option 2: Automated flow - let AI create the video draft without the middle step of adjusting
✅ Video cart in search eases quick add and adjust videos
6/8 users
user like splitting the cart and storyboard to separate rough-cut and refinement stages, so they can focus on each step.
✅ AI automation is the key
6/8 users
users prefer a prompt box to guide AI regenerate storylines, bringing more clarity to guided automation.
Design Decision - Iteration & Refinement
Iterations - Clip Panel
Iterations - Voice Edit Interactions
Before
After
I learned how critical it is to align AI features with users’ real workflows—our initial script-to-video tool missed the mark because it didn’t match creators' expectations. Through research, I discovered users preferred leveraging their existing video libraries, which led to a more impactful pivot toward an AI-assisted video-to-video tool. This experience sharpened my ability to design around user behavior and strike a balance between automation and creative control.
If I were to revisit the project, I would…
If I were to revisit the project, I’d double down on early prototyping and real-time feedback loops—not because we missed anything, but because our later iterations proved how powerful rapid user input can be. I'd also build on the personalization angle we uncovered, exploring even smarter AI suggestions tailored to each creator’s unique style. These are areas where we saw real momentum, and I’d love to push them even further.