The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Solo-built a production field-ops SaaS (TanStack Start, Postgres/Neon, Better Auth) for 275 employees — solved real-world problems like buddy punching, GPS spoofing, and devices without biometrics. Full admin + worker interfaces, live GPS tracking, still in daily use
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Trending
retrofuturism
Share your favorite designs from the past exploring the distant future. Or your own take on the aesthetic 🚀
halftone
Dots. Texture. Grain. Let's see your designs using the halftone technique.
Midjourney
Creatives are showing off what they're making with Midjourney.
colorgrade
Video editors sound off: How do you enhance the mood of your video in color grading?
Related posts
Why are AI and SaaS such a powerful combination?
AI can make a SaaS product much more useful by taking care of tasks that normally require manual work.
Some common examples:
Automating repetitive workflows
Using data to support better decisions
Personalizing the user experience
Improving security and detecting unusual activity
The key is not simply adding AI to a SaaS product. It’s about using AI where it can make the product faster, smarter, and easier to use.
Built a 3D model viewer inside @Rive .
Not a plugin, not an embed — an actual glTF renderer running on @Rive 's GPU Canvas, written in Luau and WGSL, living inside the .riv file.
Drop in a .glb and it loads, lights, animates and orbits. Every control is a data-bound property, so the whole scene is drivable from a state machine, a timeline, or app code. Nothing is hardcoded to one model.
What's in it:
• glTF parsing written from scratch — Luau has no JSON library, so the decoder is hand-rolled
• PBR shading (Cook-Torrance GGX) with key/fill/rim lighting
• Skeletal skinning, 128-joint palette
• Rigid node animation with full keyframe interpolation
• Orbit camera with per-axis locks
• Auto-fit, so any model frames correctly regardless of authored scale
• ~40 data-bound properties: camera, transform, material, animation
Most of it was built by driving the Rive editor over its MCP server with Claude Code — scripted, not clicked.
Runtime support is nearly there. GPU Canvas landed in the web runtime last week. A couple of scripting APIs still need to follow before this can ship to a live site.
Here is the Github Repo for Instructions: https://lnkd.in/dAMXQJBR
the 3D viewer is so cool. nice work!
1,000+ permissioned actions. 300+ users. Zero "who can do what" spreadsheets.
Own the Podium manages sports funding for national sports organisations, and the hardest part of the platform wasn't the funding logic — it was that permissions change depending on which stage a file is in. A reviewer can edit in stage 2 and only comment in stage 3.
We modelled access as (role × stage × action) instead of role alone. Boring decision. It's the reason adding a new role or a new stage has never meant touching the access-control code.
