Pagent

Inspiration

Everyone has a business card — something that represents who they are professionally. But sharing them is passive. You send a card and hope someone reads it.

We wanted to flip that.

What if your card could talk? What if, instead of a static card, you had an AI that knew everything about you and could answer questions on your behalf?

What It Does

Pagent turns your business card into an AI.

Upload your files — resume, deck, portfolio — and get a shareable personal page with an embedded AI agent that knows everything about you. Visitors don’t just browse your page; they interact with it.

They can ask about your experience, services, or expertise.

Share the link anywhere. It’s your business card — but alive.

How We Built It

Frontend: React + Next.js Personal profile page and chat interface

AI Models: OpenRouter Powers specialized agents generated from processed content

Database & Auth: Supabase Stores profiles, API keys, sessions, and integration tokens

Integrations:

Notion: Connect your workspace to enrich the agent’s knowledge base
Tinyfish: Data retrieval for onboarding new users/professionals
Obsidian: Sync your “second brain” with Pagent

Challenges We Ran Into

Making each generated page feel personal and unique despite wildly different input types
Keeping latency low when processing large files through sequential agents
Building the Notion integration — complex, but ultimately successful

Accomplishments We're Proud Of

Fully optimized pSEO and GEO for better LLM crawlability and retrieval
Google Lighthouse optimization for speed, reducing website bounce rates through highly efficient code and performance tuning
Multi-format ingestion (PDF, Word, PowerPoint, images) feeding into one coherent agent
Notion integration enabling richer, external knowledge beyond uploaded files
The core concept: transforming a passive business card into a conversational identity
Using the knowledge base to dynamically generate custom pages
Marketplace integration allowing developers to build modular components
Integration of i18n for global scalability, with the potential to reach 8 billion users

What We Learned

Users don’t want to think about “agents” — they just want something that works and sounds like them
Product-market fit clicked when positioned as a business card, not a chatbot

What’s Next for Pagent

More input sources: Gmail, LinkedIn, calendar — beyond just files
Agent integration: Zalo number, Whatsapp number
Agent personas: Customize tone and personality
Analytics: Track visitors, questions asked, and run A/B tests
Custom domains: Subdomains + full custom domain support
Team pages: Shared agents for agencies and small businesses
Per-user AWS infrastructure: Each user gets their own S3 bucket and dedicated agent instance (EC2/Fargate) for GDPR compliance, data isolation, and right-to-erasure — no shared storage