Lighthouz AI

Tl;dr: 500,000 procurement professionals spend their entire day manually entering data on spreadsheets and outdated software. Businesses lose $40B annually on procurement tasks that are now fully automatable. We fix this by building AI agents that automate procurement for manufacturers.

—

Hi YC, we’re Srijan and Sonali, the founders of Lighthouz AI! 👋

Srijan is an award-winning AI scientist with 60+ publications in AI and 6,000 citations. He has been a CS professor at Georgia Tech, an ex-research scientist at Google, did his postdoc at Stanford University, and PhD at the University of Maryland.

Sonali is a former lead AI engineer at Progressive Insurance, American Family, and Halliburton. She specializes in unstructured data and building AI systems from the ground up. She did her MS at the University of Washington.

We did our undergrad together at IIT a decade ago and have been married to each other for 3.5 years.

🚨 Problem: Procurement is manual 🤮

Manufacturers waste $40B annually on manual procurement processes that can be now be automated with AI. Even today, procurement teams face critical challenges:

• Manual process: 70% of procurement teams' time is wasted on manual, repetitive tasks, like verifying and entering documents on spreadsheets/ERP, comparing quotations, creating purchase orders, and communicating with suppliers. This leaves them little room for strategic negotiation activities.
• Reliance on outdated tools: 50% of procurement teams’ work is still done on emails, spreadsheets, and 90s-looking ERP systems.
• Messy unstructured data: A staggering 95% of procurement data is unstructured (think emails, scanned docs, PDFs, spreadsheets)
• The only way to scale is to hire: There is a notable deficiency in advanced tools necessary to scale operations efficiently without expanding headcount.

These challenges make the current procurement processes slow, error-prone, and lack transparency, which hinders dynamic and responsive decision-making.

🎉 Solution: AI Agents for Autonomous Procurement

Lighthouz introduces Pam, the AI Procurement specialist, who can autonomously handle manual procurement workflows and keep the human in the loop to ensure 100% accuracy.

Pam is a multi-AI agent system customized to do procurement. Pam is powered by dynamic knowledge graphs extracted from unstructured data, personalized for each business-supplier relationship. Pam integrates with your email and ERP systems. Pam provides an intuitive chat-based UI for procurement teams to be productive immediately.

Pam helps procurement teams by:

✅ Automatically entering data on ERP systems, saving countless hours every week

✅ Offload tedious activities of quote comparisons and vendor recommendations

✅ Creates purchase orders from quotes automatically

✅ Flag errors in quotes and order acknowledgments to prevent delays

✅ Spot overbilling in invoices to prevent revenue leakage

Advantages of using Pam 🔥

✅ Pam works 24/7 on multiple orders

✅ You can scale your procurement team with Pam at a fraction of the cost without increasing headcount

How can you help ❤️

1. Connect us to CEOs of mid-market manufacturing companies in your network. A quick forwardable blurb: “A YC-funded startup from Stanford, UW, and Google has built an AI-powered procurement system, which helps procure faster, increase productivity, and increase cost savings. You can reach out to srijan@lighthouz.ai to discuss further.”

2. If you’re working in or have worked in procurement for manufacturers, we’d love to connect and chat. Send us a message at srijan@lighthouz.ai 📨