Vyra
Co-founding an AI SaaS that streamlines authentic scriptwriting and short-form video editing for marketers.
Overview: Automating Short-Form Content
Vyra is an AI tool that turns raw ideas into viral video scripts and video content. As Co-founder and Lead Product Designer, I turned my own agency’s manual work into an automated platform. I designed the brand, the website, and the tool itself to make professional content creation fast and simple.
Role: Co-founder & Lead Product Designer
Timeline: November 2025 – Present
Team: Caleb Pong (Co-founder & CTO)
Disciplines: 0→1 Product Strategy, AI UX, Brand Identity, Marketing
Tools: Figma, Framer, Vercel, Railway, Supabase, Modal
Problem: Content Creation is Time-Consuming and Resource-Intensive
82% of marketers struggle to create high-quality content consistently [HubSpot 2024], with the average short-form video taking 7–10 hours to produce manually [Statista 2024]. Standard AI tools have not solved this; they often produce robotic scripts that fail to engage audiences. Through 50+ user interviews, I found that business owners were frustrated by the "inauthentic and ineffective" scripts generated by basic LLMs and remained confused by the complexities of video editing.
System Flow: Replacing Complex Timelines with Natural Language
I designed the user journey to bridge the gap between a raw idea and a production-ready video. The flow centers on two primary actions: Authentic Scripting and Agentic Editing. By implementing a natural language interface, I enabled users to edit video using English commands instead of manual timelines. This system utilizes a personalized memory of the user’s voice and project history, allowing the interface to adapt to individual styles and reduce repetitive prompting.
Execution: Founding a Technical Startup
I built a minimalist dashboard that automates the technical heavy lifting of trend-tracking and video assembly. To solve the "robotic AI" issue, we trained our models on proprietary viral datasets, resulting in scripts that were 90% more aligned with high-performing content than standard LLMs. By integrating the Gemini 3 Flash API with Modal for serverless GPU compute, we ensured the platform could handle complex rendering tasks while maintaining a low-latency user experience.
We are currently gaining traction with our first group of clients, partnering with startups to launch their brands while simultaneously scaling our own content presence with Vyra-assisted content.
The Outcome: Achieved Market Validation and Operational Efficiency
Efficiency: 10-hour production cycles for a short form video reduced to 10-30 minutes.
Traction: 100+ person waitlist with active startup client onboarding. Paid traction with a done-for-you marketing service, using Vyra internally.
Quality: 90% higher script authenticity compared to standard LLM outputs.
Retrospective: Building for the Future of Agentic Workflows
Natural Language UI: Replacing technical complexity with simple English commands to lower the barrier for non-editors.
Personalized Agentic AI: Prioritizing user history and "voice" over generic data to maintain authentic output.
Work With People You Love: So that when things are inevitably difficult, everything seems just a little less bad.
Dual-Track Growth: Leveraging an internal agency tool to accelerate SaaS iteration and user empathy.
Next Steps: Scaling the agentic engine to support multi-platform distribution and automated content scheduling.
