No Borders

Inspiration

As immigrants from the global south who came to the UK, my co-founders and I struggled to navigate the complex visa process without sufficient support. The visa officers at our company meant well but lacked the expertise and incentive to guide us to the optimal visa route. We wasted thousands of pounds applying for visas that didn't fit our circumstances, like the Skilled Worker visa, when we actually qualified for more affordable options like the Global Talent visa.

This frustrating experience revealed a major gap in quality visa assistance for immigrants like us. We realized that many newcomers to the UK, especially those with limited financial means, are equally lost when it comes to securing the right visa. They deserve an advocate who deeply understands the system and will help them identify the most suitable visa in a cost-effective way.

That sparked our idea to create a visa support service tailored to the needs of global talent coming to the UK. Our goal is to make the visa process smooth, stress-free and financially sensible for our users. With the right guidance, immigrants can avoid costly mistakes and focus on fully realizing their potential here. We want to empower global talent to build lives and careers in the UK with confidence

What it does

Our visa assistant chatbot is specially trained to guide users through the UK visa process from start to finish. It has an in-depth understanding of the latest official visa guidelines and requirements. This allows it to respond accurately to any visa-related question and provide tailored recommendations based on the user's unique situation.

The chat interface makes querying our bot intuitive and conversational. Users can ask follow-up questions to clarify or expand on a topic, with the bot able to reference earlier context in the chat history. Its knowledge encompasses the full spectrum of UK visas so it can point users to the optimal visa route.

How we built it

Our solution focused on entry to the UK with regard to the four most popular visa applications. These are Global talent, skilled worker, healthcare and graduate visa. We scraped the available documentation on the UK GOV websites for each of these visas and put these into PDFs, after cleaning and removing erroneous tags in the documents. Once done we used the ParentDocument Retriever in langchain to store these into Pinecone and the InStoreMemory in LangChain.

From this point we used the LangChain ConversationalQA Chain to interrogate the Vector DB using CLAUDE 2 in the backend.

For the front end, we used Streamlit to expose a chat style interface and also built in functionality to display the sources of the information (eg which document, which chunk, which page).

Challenges we ran into

Finding all of the requisite information for applying for a particular visa.

Accomplishments that we're proud of

Creating an interface which provides more detailed and factual information about UK visas than CLAUDE 2.

What we learned

The minutia involved with applying for UK visas means there is a great deal of detail and complex queries that exist. Browsing online forums such as Reddit focused on UK visas, presented a range of very nuanced use cases where having a centralised VISA assistant would be very helpful.

With regard to the technical front, learning about the prompt engineering for CLAUDE 2 thanks to the workshops available throughout the Hackathon has been excellent.

What's next for No Borders

Enhance it with the complete suite of UK visa documentation Create a front end with more functionality (filtering, document upload) Enhance it with user entry section where the users populate their details and the LLM can automatically suggest the visa required for a chosen country.

Built With

• claude
• langchain
• pinecone
• python
• streamlit