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
Built and deployed TechNova RAG AI Assistant — an intelligent document-based assistant that lets users ask questions about their documents and receive relevant, source-backed answers.
The assistant uses Retrieval-Augmented Generation (RAG) to retrieve relevant information from uploaded documents before generating responses, helping reduce unsupported answers and making document search more useful.
Key features:
• Document-based question answering
• RAG pipeline for contextual responses
• Source references with answers
• Conversation memory
• Casual English and Roman Urdu conversations
• AI assistant-style chat interface
• Streamlit deployment
Tech Stack:
Python • RAG • LLMs • LangChain • Vector Database • Streamlit
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
The Assumption of the Virgin, Bernardo Daddi, 1337–38. Tempera on wood, gold ground. The Met, Robert Lehman Collection (1975.1.58).
This panel is the top half of an altarpiece for a chapel in Prato Cathedral, built to house the city's real relic: the Virgin's girdle. In the scene she lowers it to Saint Thomas below as proof her body rose whole – the panel showing his hands catching it is lost.
We rebuilt it as a tiara. The mandorla becomes the crown's pierced opening. The angels return in glass, still in their original green and pink. A frog takes the summit – the one animal that changes bodies without dying. A poppy blooms where the Virgin sat: not a relic passed down to a man waiting below, but a small, symbolic dawn of the matriarchy still ahead of us. 🌸 💅
Original Met piece: metmuseum.org/art/collection/search/458998
Inst: https://www.instagram.com/reel/Dc8y5NMIivZ/?stkn=NHlrcXg3dTlkemJl
Video experiments: https://app.flora.ai/join-project/04c3dfff-e5db-4819-afd6-54c7a7bf7725
Your AI agent just did 9 things: pulled 412 orders, matched them to Stripe, caught a duplicate £248 charge, drafted a ledger entry, opened a ticket. Whole run cost $0.011.
Now the real design question — what does the user actually SEE?
A — The transcript. A paragraph of text and a collapsed "Used 3 tools".
B — The run canvas. Nine colour-coded rows: what ran, what it cost, what got flagged, and one step that stops and waits for you before money moves.
Same agent. Same run. Two completely different products.
Vote A or B — and tell me why. I build the B version for a living.
0 voted
1%
0 voted
1%
0 votes
Closed
I built a workflow that turns a business website into a testable AI phone receptionist.
The system analyzes the website, extracts services, staff, opening hours, policies, and other relevant information, then creates an isolated voice-agent demo for that business. Instead of sending a generic recording, we can provide a real phone number that allows the business to test the agent through an actual call.
When the website or scheduling system supports an integration, the agent can check availability and book appointments during the call. Otherwise, it safely captures the caller’s details and preferred time for staff follow-up.
If a business decides to continue after the demo, the same foundation can be expanded into a production receptionist with databases, customer-specific business rules, calendar and CRM integrations, confirmations, call summaries, analytics, and administrative controls.
The extracted information can also reveal gaps in the business website. Its content and structure can then be improved so both visitors and the AI receptionist receive clearer, more reliable information.
The main lesson I share is that website extraction alone is not enough. Information must be normalized, validated, and restricted to the source material before it can be used safely in a live voice conversation.
The demonstration uses a fictional dental clinic and real recordings from my working voice-agent system. No patient data, real customer information, or invented testimonials are included.
I used Descript to assemble and refine the presentation, edit the narration and real-call sections, create captions, and prepare the final submission.
