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
Overview
DocuMind AI is an enterprise-grade multimodal document intelligence platform designed to extract, analyze, and synthesize critical insights from messy unstructured PDFs, financial statements, and legal agreements in real time.
The Problem
Traditional OCR and text extractors lose contextual hierarchy, choke on multi-column tables/scans, and induce heavy latency—often freezing the UI while large documents are processed. Moreover, sending raw full documents to LLMs causes expensive token overhead and hallucinations.
The Solution & Architecture
Multimodal Ingestion Pipeline: Streamed document parsing using Claude 3.5 & Gemini Multimodal models for rapid vision-level layout extraction and risk identification.
Dynamic Context Compaction: Built intelligent chunking and token-budget pruning algorithms, reducing LLM API token consumption by 42% while eliminating hallucinated clauses.
Sub-Second Streaming UI: Implemented Next.js 15 Server-Sent Events (SSE) streaming with a dual-pane responsive viewer (original document on the left, interactive citation breakdown and risk matrix on the right) with zero main-thread lag.
Key Impact & Metrics
99.2% Accuracy in extracting high-risk contractual clauses and numerical tables.
< 1.2s Time-to-First-Token (TTFT) on 50+ page PDF documents.
40%+ Cost Reduction via aggressive prompt caching and token pruning.
The split view makes sense.
Thanks Vansita! That was the idea: keep the source text right next to what the AI found. Have you worked with document review tools before? Would love to hear what you'd want to see in something like this.
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
Related posts
Crenaly: a booking SaaS I designed and built end-to-end for appointment-based businesses in France: salons, barbers, beauty and wellness.
The core idea: when a client cancels, the freed slot is automatically offered to clients who asked for an earlier time. The first to accept gets it, their appointment moves in place, and the slot they leave is offered next. No double booking, no manual work.
Around it, a full product:
→ Branded booking page for each business, embeddable on their own site
→ Mobile-first booking flow, no account needed
→ Agenda, client management, team roles & permissions
→ Verified reviews, analytics, Stripe billing
Product, UX/UI, frontend and backend.
Next.js · TypeScript · NestJS · PostgreSQL · Redis
Automatically offering a cancelled slot to the next client is such a practical feature for salons and barbers. Designing and building Crenaly end to end must have made those flows much tighter.
Most AI apps work like this: the model decides, the app saves, and the user finds out later that the numbers are wrong. 😬
For Notarik, I set one rule. The AI can only propose. Nothing gets saved until the user confirms. ✅
🧾 What it does
Take a photo of a receipt, or type "lunch 45k at the warung". You get a categorized expense, a monthly recap, and a chat assistant you can ask things like "where did my money go this month?":
🔒 The parts I care about most
• Receipt totals are checked by plain code, not by the model. If the math doesn't add up, it gets flagged.
• Money is stored as whole numbers (IDR), never decimals, so there are no rounding bugs.
• Every user's data is locked at the database level with Postgres row-level security. Even if the app has a bug, it can't show someone else's records.
• Text on a receipt is treated as untrusted. If a receipt says "ignore your instructions", it's just text.
• The agents have tests, so when I change a prompt I can see right away if it got worse.
🛠 Built with Next.js, React Native, Python (FastAPI), LangGraph and PostgreSQL. One person, web and mobile, from design to API.
🤝 If you're building something where an AI touches real money, records or user data, this is the kind of work I like: agents that are useful and also safe to put in front of users.
👉 Try it live: https://notarik-app.nashir.tech
Want to talk about yours? Send me a message.
Smart call keeping the confirmation.
Quote Follow-Up Automation for HVAC & Plumbing
Most home-service quotes don’t get rejected, they just go quiet. I built a system that follows up on them automatically.
What it does
Takes each quote through a simple form
Schedules up to 8 timed texts over 30 days
Stops instantly on a reply, a STOP, or a booked/lost update
Uses AI to draft replies to objections, with a human approving every one
Built with safety first
No duplicate texts, even when processes overlap
Only contractor-approved wording is sent
Texts go out only during local business hours
Opt-outs are honored immediately
Stack: n8n, Postgres, Groq, Docker, Twilio
Status: built and tested end to end, ready to pilot.
I build automation systems like this for service businesses. Message me if you’d like one for yours😊 .
