Nim is a full-cycle AI video production studio - a web app where freelancers and creative teams can make medium-length AI videos 10x faster

At a glance

From creator to founder

In Fall of 2023, I started making AI videos as a personal hobby Made at least one AI-made post on TikTok every day for three months Spent hundreds of hours in Midjourney, Runway, Pika, ElevenLabs, Stable Diffusion, ComfyUI

Creating AI videos was too time-consuming Anything medium-length (brand video, music video, short film) took over 100 hours of work ****All existing AI video apps focus on individual shot creation (1-4 sec video clips) Most of my time costs were in other parts of the process **** Find references for inspiration Find images, 3D scenes, and other reusable assets for AI transformation Develop storyboard and write hundreds of individual prompts Find types of shots that AI can do well enough Try to improve shot quality in post-production Switching between at least 7 apps to make a single video

Market gap: creators need full-cycle workflow app to make production 10x faster 1M+ tried AI video tools, but less than 1k actually delivered a single paid AI video project Reducing time cost to 5-10h would dramatically grow the number of creators and applications Faster creation → more creators → more training data → better AI → more applications

Getting started Raised $700k in early February to build a complete AI video production studio SignalFire, who previously funded video editing unicorn Frame.io, backed Nim on day one

Building the app I wanted as a user

Core use case discovered: social video ads for consumer brands Agencies want new revenue Brands want to stand out on Instagram, TikTok, Youtube Shorts and are ready to pay Current AI is good enough to make high-engagement short videos for social media One problem: with 100h+ time cost, AI projects are unprofitable for agencies

Solving productivity challenge with Nim 8 AI tools, capable of making shots like these: Nim showreel Group actions for fast upscaling and image-to-video tasks Explore section to find public works for inspiration and reuse Showreel to share best shots with friends and clients Creative assistant: a GitHub Copilot-like chat to help with prompts and storyboard

Open source foundations and specialized training in-house Nim is built on open-source foundational models like Stable Diffusion and DynamiCrafter We continuously track new open source models and add them to the platform Commercial third-party models like ElevenLabs can be added via APIs in the future We train specialized models for tasks that aren’t solved well by open source Motion control, video style transfer, regional edit + more in progress

Demo