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
Support teams answer the same questions all day, and generic chatbots guess. I built an n8n AI agent that answers only from the business's own documents. New files in Google Drive are downloaded, embedded with Google Gemini and stored in a Pinecone vector store (307 items in this build). A chat-triggered agent with memory searches that knowledge base and replies with accurate, context-aware answers. The same setup extends to email, Slack, Shopify or CRM support, with human handoff for tricky cases.
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
Landing page for an AI-native terminal platform
Most lead generation tools collect leads.
Mine scores them, analyzes them, and logs them automatically.
I built an AI Lead Intelligence Engine using n8n that triggers the moment a form is submitted. It normalizes the data, runs it through an AI scoring model, detects industry and buying intent using Gemini, assigns a HOT/WARM/COLD tier, and saves everything to a live CRM sheet.
Built for businesses that are tired of chasing cold leads manually.
This is Stage 1. Stage 2 adds automated personalized outreach. Stage 3 adds an AI voice call within 60 seconds.
If you run a business and lose leads because follow-up is slow, this system fixes that.
Love this Lead Intelligence flow! I also build lead scoring with n8n + OpenAI + Google Sheets. Clean execution — how do you handle duplicate leads? Let's connect!
🟢 I built an AI agent that reads invoices so humans don't have to. Manual invoice entry is slow, boring, and full of typos. So I automated it end to end in n8n.⚫
Here's how it works:
🟢 Invoice comes in: A webhook receives the PDF through a POST request
⚫ Text extraction: The file is parsed and the text pulled out
🟢 Safety check: If the file is invalid, the system responds with an error right away
⚫ Smart chunking: Large invoices are split into smaller pieces so nothing gets cut off or lost
🟢 AI extraction: A loop sends each chunk to an Information Extractor powered by Anthropics Claude, which pulls out the fields the client needs (vendor, invoice number, dates, line items, totals)
⚫ Merge and validate: The results are combined into one record and checked before they are sent
🟢 Clean JSON response: The structured data goes straight back through the webhook The hardest part was'nt the AI. It was making it production-ready. 💡
Long, multi-page invoices break most simple AI workflows because they exceed the models limits. Chunking, looping, merging, and validating are what let this handle large invoices reliably. Where this fits in the real world:
⚫ Accounting and finance teams processing hundreds of vendor invoices
🟢 Businesses syncing invoice data into ERPs or spreadsheets
⚫ Freelancers and agencies automating their billing workflows What used to take minutes per invoice now takes seconds, with no copy-pasting. ⚡ I'm currently going deeper into n8n and AI automation.
If you are looking to automate document processing or any repetitive workflow, lets connect! 🤝
#n8n #Automation #AIAutomation #InvoiceProcessing #WorkflowAutomation #Claude #FullStackDeveloper #NoCode
