Pan'l

• 

  Discover how Pan'l transforms web content into structured data with our advanced AI technology - reusable components for API services

• 

  Transform HTML content into structured, reusable components that fit perfectly into your workflow.

• 

  Connect with your essential tools and services with a single click. Build powerful workflows that span your entire tech stack.

• 

  Multiple ways to select exactly the data you need from any webpage with precision and ease.

• 

  Monitor your data extraction activities with real-time analytics. Track token usage, analyze performance, and optimize your workflows.

• 

  Embed Pan'l on any website instantly

Inspiration

Inspired by trees :) and the people gravitating around them

What it does

The project is a JS widget that allows any website to offer its users an AI tool to quickly transfer data to services such as Slack, Hubspot, Google Sheet, Drive, Docs, etc. (Thanks to Pica for the 13,000 connected service APIs)

Through a page scan, manually drawn area, precise selection, or voice request, the user chooses which data they want to use. The AI then analyzes this content to create "reusable components". The user selects the interesting components, chooses the services they want to work with, and asks the AI what action to perform on the selected services.

For Example:

Step 1: Voice request: "Select the contact table on the page" (STT Elevenlabs)

Step 2: The user selects the table from the component(s) generated by the AI

Step 3: The user connects their Hubspot service

Step 4: Voice request: "using the table, create contacts on Hubspot"

=> The AI creates contacts on Hubspot from the current page, without having to go to Hubspot and without effort :)


How we built it

The widget, API, app, demo, and marketing site were all initially created with Bolt.new, then reworked with VSCode and Cursor.

Challenges we ran into

The AI features had to be fine-tuned multiple times to achieve the best possible results. They still need improvement, and this is Pan-l's next challenge to offer the most intuitive and simple experience possible. Will need to continue improving the analysis, components, and communication with services for an even better user experience.

Accomplishments that we're proud of

The small green leaf in the top right corner of the widget. The leaf informs the user of the estimated consumption in Wh or Kwh and the lighting time of a 10W bulb for each AI request. It's barely visible, but it has important meaning for the project's future.

What we learned

• Vercel AI SDK and communication with different LLMs
• Pica
• Supabase
• Next.JS (already worked with React in the past though)

What's next for Pan'l

Looking for enthusiastic partners to shape the future of this project.

Built With

• bolt.new
• brevo
• claude
• clerk.com
• elevenlabs.io
• gemini
• mistral
• next.js
• node.js
• ollama
• openai
• picaos.com
• stripe
• supabase