Tendra is a web app I designed and built for people who make video with AI. Instead of juggling a separate account and interface for each model, you connect your own provider keys once and run everything from one place.
You write a prompt once and run it across models such as Veo, Kling, Sora and LTX-Video side by side, then compare the output, cost and render time of each. If you would rather not choose, you set a priority (quality, price or speed) and Tendra scores the eligible models and sends the job to the best one. You can also vote on anonymous A/B renders, so the app learns which models you actually prefer.
Your provider keys live in an encrypted vault. They are encrypted at rest with AES-256-GCM, decrypted only on the server for your renders, and never shown again. Compute is billed by the provider directly to you, with no markup.
Every render is copied to a permanent library, so your links don't expire when the provider's do, and you can trim, crop and stitch clips in the browser. For talking videos there is a Recut studio, where you edit the video by editing its transcript: delete a sentence and the video re-cuts to match.
I handled the product design, front end, back end and the promo video. The video above is built in code with Remotion, using real screenshots of the app with the camera moves, cursor and interactions animated in React. The app itself is built with Next.js, React, TypeScript, Supabase and Redis.


0 likes · 6 views · Jan 24, 2026
Tendra: one studio for every AI video model Tendra is a web app I designed and built for people who make video with AI. Instead of juggling a separate account