Inspiration

One of our teammates has a close friend who runs a small manufacturing business and wanted to start exporting his products abroad. He hit a wall almost immediately — every time he searched for how to classify his product under the right HS code or what paperwork he'd need, he found scattered, contradictory information spread across government PDFs, forums, and outdated blog posts. The alternative was hiring a customs broker, and the fees quoted to him were more than his shipment margins could absorb. He's not alone — most first-time SME exporters, especially in emerging markets, don't have an in-house legal or customs team and can't afford one. That conversation is where TradeEase started: could we build something that gives a small exporter the same clarity a paid consultant would, without the price tag?

What it does

At present, it takes input text data from user and use this in our three modes.

  • HS Code Classification — describe your product in plain language and get a suggested tariff code, an explanation of why it fits (using WCO classification logic), a confidence rating, ask clarifying questions if when confidence falls below a certain level, and a flag for anything you should double-check with a licensed broker.
  • Commercial Invoice Generation — turn shipment details (exporter/buyer info, product, Incoterm, quantity, pricing) into a properly formatted, export-ready invoice draft.
  • Guidance & Mentor chat — a quick-question chatbot for the everyday "wait, what's an Incoterm" or "do I qualify for AGOA" moments, instead of opening five browser tabs.

How we built it

We started with a PySide6 desktop prototype wired to Google's Gemini API (paid tier 1) so we could iterate on the UX and prompts quickly, then we used streamlit library to deploy it as a web app.

Challenges we ran into

We are a new team of 4 members who don't have much experience of hackathons , our biggest challenge was efficiently dividing the project work. Also other hard problem was scope. It was tempting to build a full "personalized action plan" generator and "more documents generator" as part of the guidance feature, but we made the call to keep it to a focused Q&A mentor and only invoice generator for the MVP and clearly mark the bigger version as a roadmap item instead of half-building it. We'd rather ship three things that work than four things that are fragile.

Accomplishments that we're proud of

Getting a real, working pipeline from desktop prototype to a deployed web app in the hackathon timeframe, end to end, on mostly free infrastructure. We're also proud of where we drew the line on AI trust — the caution fields and human-in-the-loop framing weren't an afterthought, they're baked into the prompts themselves, because we kept coming back to the fact that this is a tool for someone's real shipment, not a toy.

What we learned

We went in knowing nothing about international trades to some portion of how international trades work, we learned about the importance of hs codes classification and gri rules ,many different bodies are involved in international trades. Also how much important are the SMEs for a country's economy and job sector. On technical part, we learned about importance of good prompts and using gemini through api key and setting up gemini-genai-sdk and also learned about pyside6 library and streamlit library.

What's next for TradeEase

It is a highly ambitious project trying to provide end to end support for SMEs, beginning with hs code classifier, invoice draft generator and Q&A AI mentor. In future it will include these features on minimum subscription fee-

  1. Multi-Document Consistency Engine-- ● Automatically compiles the core export paperwork queue simultaneously: Commercial Invoice, Packing List, and Certificate of Origin. ● Treats documents as a unified ecosystem rather than separate forms. The AI ensures names, weights, and legal descriptions match perfectly across all files to eliminate border friction. 2.Proactive Tariff Optimization-- ● Simulates trade routes actively. If a route qualifies for regional trade agreements (e.g., South Africa to the USA), the system triggers a critical dashboard alert: “This product qualifies for duty-free access under AGOA. Ensure you file Form 350 to save an estimated 12% on entry tariffs.” ● Pairs generated documents with a step-by-step port execution plan.
  2. Data of different buyers and sellers, and matching them automatically based on their profile and what they need.
  3. Soon it will be multilingual, more localised, and multimodal AI.

Built With

  • chatgpt
  • claude-code
  • claude-sonnet-4.6
  • gemini-3.5-flash
  • gemini-api
  • gemini-genai-sdk
  • git
  • github
  • pyside6-library
  • python
  • streamlit-framework
Share this project:

Updates

Submission history