EthosLens

Ethical shopping simplified! Our AI researches & scores brands on sustainability & labor. Get transparent reports for confident, ethical choices.

Comment

Inspiration

We noticed a growing frustration among consumers: 74% find it incredibly difficult to get clear, trustworthy information about how ethical brands really are. People want to support companies that do good – treating workers fairly, protecting the environment, and being transparent – but the truth is often buried in long reports or hidden by confusing marketing. This inspired us to build EthosLens, a tool to make ethical information easy to understand and accessible to everyone.

What it does

EthosLens is an AI-powered research assistant that automatically investigates a brand's ethical practices. You give it a brand name, and it:

  1. Searches the internet for relevant documents like sustainability reports and news articles.
  2. Reads and understands these documents (even complex PDFs).
  3. Generates a simple report summarizing the brand's ethical performance.
  4. Provides an overall ethical score, broken down into key areas like environmental impact and labor practices, with evidence from the documents.

How we built it

We used a combination of powerful tools and Python libraries:

  • Python: The main programming language for the entire backend.
  • Google Custom Search API: To find relevant documents online.
  • Unstructured.io: To read and clean up different document types (like PDFs and web pages).
  • ChromaDB: A special database to store and quickly search through the text information.
  • LangChain: To manage the flow of information and communication with the AI model.
  • OpenAI GPT Models: The AI "brain" that reads the text, answers questions, and helps calculate the scores.
  • FastAPI: To create a web API so other applications (like a website) can use EthosLens.
  • Asyncio: To make many parts of the process run at the same time, speeding things up.

Challenges we ran into

  • Messy Data: Real-world company reports and web pages can be very disorganized. Getting the AI to understand them correctly was tricky.
  • Information Overload: Sometimes we'd find hundreds of documents. Teaching the AI to pick out the most important information for scoring was a challenge.
  • Speed: Reading and analyzing so much information can be slow. We had to work hard to make the process faster using asyncio for parallel processing.
  • AI Consistency: Sometimes the AI wouldn't follow instructions perfectly for scoring. We had to carefully adjust our prompts to get reliable results.

Accomplishments that we're proud of

  • Fully Automated Reporting: We built a system that can go from just a brand name to a detailed ethical report with scores, all automatically.
  • Handling Complex Documents: EthosLens can process and understand complicated PDF reports and messy web pages, which is a tough problem.
  • Evidence-Based Scores: Every part of the score and report is backed by direct quotes and links to the original documents, so users can trust the information.
  • Fast Analysis: By making different parts of the system run at the same time (like searching, downloading, and analyzing), we made the whole process much quicker.

What we learned

  • AI Needs Good Data: The quality of the AI's analysis depends heavily on the quality and relevance of the information it's given.
  • Breaking Down Problems: Complex tasks, like ethical analysis, are easier to solve by breaking them into smaller, manageable steps.
  • Speed Matters: Making software fast and responsive is key for a good user experience, and tools like asyncio are very helpful for this.
  • Clear Instructions for AI: AI models need very specific and clear instructions (prompts) to perform tasks reliably, especially when you need structured output like scores.

What's next for EthosLens

  • User-Friendly Website: Create an easy-to-use website where anyone can type in a brand and see its ethical report.
  • More Data Sources: Include information from social media, news sentiment, and customer reviews to get an even broader picture.
  • Track Changes Over Time: Allow users to see if a brand's ethical performance is improving or declining over months or years.
  • Compare Brands: Let users easily compare the ethical scores of different brands side-by-side.
  • Deeper Analysis: Improve the AI to better understand nuances in company reports and identify potential "greenwashing" more effectively.

Built With

  • chroma
  • fastapi
  • google-programmable-search-json-api
  • gpt-4.1-mini
  • langchain
  • next.js
  • openaiembeddings
  • react
  • tailwindcss
  • unstuctured
Share this project:

Updates