Agentic Experience Design: Building an AI Agent Inside a Live Healthcare Dashboard The Problem In...Agentic Experience Design: Building an AI Agent Inside a Live Healthcare Dashboard The Problem In...
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
Agentic Experience Design: Building an AI Agent Inside a Live Healthcare Dashboard
The Problem
Innohlth, a clinical operations platform (RRT, Code Blue, ICU transfers, patient & staff management), relied on manual dashboards, filters, and workflows. No agent. No automation. Just static screens.
The Fix
1 ) I audited every read & write operation to identify where an agentic experience could replace manual work, then designed and built the agent:
2) Mapped workflows and defined what the agent should know, do, and how it should respond
3) Cloned the existing repo, created an isolated branch, and kept production untouched
4) Used Claude Code to build, test, and refine the agent and its generative UI against the real codebase
Verified everything in a safe environment before shipping
The Result
"Ask RRT" is a fully agentic copilot embedded into the dashboard:
Understands intent through natural language not just keywords
Generates charts and visual summaries live with Generative UI
Returns structured responses with summaries, key findings, and attention areas
Integrates seamlessly without disrupting the existing platform
If you have an existing product and want a real AI agent with Generative UI, not just a chatbot designed and shipped safely into your live system, let's connect..
AI fashion editorial. I used a selfie, a photo of the Nour Hammour "Sandy" coat, and leveraged GPT Image 2.5 Sunburst + Seedance 2.5. That Nour Hammour Paris coat is on my wishlist.
I stopped making files and started making systems.
One Houdini scene, one switch: floral, stone, dragon. Every product drops into the same room, the same camera, the same light. The only thing that changes is where the switch points.
How do you handle variations, one scene or one file per product?
https://contra.com/p/cvXglUcM-product-film-retractable-shelving?r=oscar_romero_0l59be1e
One parametric scene wins every time once a client asks for a fourth variant, that's usually the moment a per-product file approach starts costing real hours. Does the switch just swap materials and geometry, or does lighting need any tweaking too when you flip between floral, stone and dragon?
Playing around with SaaS motion using Claude + After Effects + Higgsfield.
Just started experimenting with smooth movements, little bounces, and some morphing techniques.
Still figuring things out, but this workflow is pretty interesting. If this keeps working, making a full explainer video might not take a whole week anymore. 😄