An unfinished vibe-coding experiment in AI-assisted product marketing.
This repository is not a turnkey product and is not immediately adoptable as-is.
ProductMotion.app was built as an experiment in AI-assisted / "vibe coded" SaaS development rather than as a fully finished commercial product. The backend and authenticated dashboard progressed much further than the public marketing surface, but the landing page was never properly completed and the overall product was never taken through the final polish needed for production adoption.
I am not actively continuing development on this project. My main focus is on building professional tools that support my day job more directly and make that work more efficient and meaningful. Still, this remains a very dear project to me: I loved building it, I still think the underlying idea is good, and I think someone else could take it further.
If you want to adopt it, the intended path is probably to fork it, modernise it, replace older model choices where appropriate, and finish the product experience.
ProductMotion.app started as a very specific response to a new capability wave in AI image tooling.
When Google's original Gemini 2.5 Flash Image model — widely referred to as Nano Banana — was released, it immediately opened up a new category of lightweight product ideas. Around that moment, I was particularly inspired by Pieter Levels (@levelsio), whose posts often highlight how new model capabilities can become narrow, fast-shipping software products.
One post in particular captured the moment well:
- Pieter Levels on X: "If you're not building a mini startup with Nano Banana today..."
That post landed right as Google officially introduced the model:
- Google Developers Blog: Introducing Gemini 2.5 Flash Image (aka nano-banana)
The idea behind ProductMotion.app was simple:
take a newly unlocked AI image capability, combine it with video generation, and package it for a very specific niche of people who sell products online and need better-looking product visuals without expensive studio workflows.
That felt commercially interesting because many online sellers, small brands, and indie operators did not have a cheap, repeatable way to generate polished product visualisations. The product concept was to turn 1-2 product images into more dynamic marketing outputs such as rotational or motion-led product videos.
So this repo became one of my early serious experiments in applying new model capabilities horizontally to a clear market need: not building a foundational model, but wrapping a new model in a product workflow for a specific buyer.
I still think that part of the idea holds up.
What exists here proves that the concept was technically viable: there is meaningful backend work, authentication, dashboard work, upload flows, and internal product logic. What does not exist is the finished public-facing product layer that would make it easy for an outsider to adopt or trust immediately.
If somebody wanted to continue it today, they would probably:
- refresh the UX to avoid the older AI-generated dashboard feel
- build a proper landing page and clearer onboarding flow
- revisit the model stack with newer image and video models
- simplify deployment and packaging for outside adopters
At a high level, this repository contains:
- a FastAPI-based backend
- authentication and user-management work
- an authenticated dashboard flow
- image/video generation experiments and supporting scripts
- extensive architecture, implementation, and debugging documentation from the build process
What is relatively strong:
- backend experimentation
- authenticated dashboard work
- technical documentation/history
- evidence that the core workflow could be built
What is still incomplete:
- public landing page
- polished product UX
- packaging for outside adopters
- production-grade end-to-end finishing
If you are interested in the underlying idea, treat this repository as:
- a technical starting point
- a documented experiment
- a fork-and-finish candidate
Do not treat it as a ready-made SaaS you can deploy unchanged.
- Public overview: this
README.md - Maintainer-oriented operational notes:
docs/MAINTAINER-RUNBOOK.md - Architecture and implementation docs:
docs/ - Agent workflow context:
AGENTS.md