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
Prepwise is a retrieval-augmented generation (RAG) system that generates exam-quality questions from a student's own course material — lecture PDFs, slides, and scanned notes.
The scanned-notes path is where quality leaks, OCR errors propagate straight into the questions with no recovery downstream. Worth noting RAG for question generation wants different retrieval than RAG for answers, you need a coherent concept span per question or you end up testing trivia from whatever chunks matched.
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
One thing I care about when building interactions like this is accessibility, especially making sure they still work well when you’re not using a mouse.
For this merchant onboarding component, that meant:
- Keyboard navigation with a logical tab order
- Focus management through expand and collapse
- ARIA states for disclosures and dropdown menus
- Clear focus states across light and dark
- Reduced motion when the user prefers it
A lot of this isn’t really obvious from looking at the UI, but it’s part of the build for me.
Reduced motion support is the one people forget most, and it's invisible in the UI but still part of the build, that framing is spot on. Most non-technical clients would never think to ask for any of this. Do you have a go-to way of testing focus order beyond just tabbing through manually?
This is one of the product website i created with Next.js, typescript etc, for my client.
This is so professional!
I designed and developed a specialized AI phone receptionist for veterinary clinics.
The system handles real PSTN and SIP calls through FreeSWITCH. It combines streaming speech recognition, LLM orchestration and low-latency voice synthesis to communicate naturally with pet owners.
The agent can answer questions about veterinary services, business hours, location, parking and clinic policies. It can identify the pet owner’s reason for calling, distinguish routine appointment requests from potentially urgent situations, check configured availability and guide the caller through a veterinary visit request.
The system includes:
• Real US telephone and SIP call handling
• Pet-aware reason-for-visit collection
• Routine, sick-visit and urgent-request routing
• Veterinary service and provider information
• Appointment booking, rescheduling and cancellation flows
• Email confirmations and owner notifications
• Safe escalation for urgent or unresolved situations
• Call recordings, transcripts and structured summaries
• Appointment status and administrative review
• Configurable services, staff, schedules, policies and instructions
• Replaceable speech recognition, LLM and voice providers
I developed the telephony architecture, real-time conversational pipeline, backend APIs, veterinary call flows, appointment logic, email integration, administration, deployment and real-call testing.
The agent does not diagnose animals or replace veterinary judgment. Urgent, uncertain or clinically sensitive requests are handled according to configured escalation rules.
This is an independently developed product, not a paid client project.
