Inspiration
John, Liam, and Alex have all worked on previous startups and came together earlier this year to create Sonder after researching opportunities to help small banks with new technology under new open banking regulatory pressures. After performing over 30 problem and idea validation interviews, the Sonder team is continuing to chip away towards helping small banks.
What it does
Sonder Junior Analyst is an agent designed to interface directly with a data analyst or loan officer within a bank. The agent is able to organize all the different data sources within a bank and make it very simple for a data analyst to do their job.
How we built it
We built the agent using a combination of tools including langchain, nextjs, sql.
Challenges we ran into
The data within a bank is extremely fragmented so the data source our team used from Jack Henry ODI exports was not only large but also very unorganized. Having the agent efficiently and accurately create requests to the database based on user queries was a challenge.
Accomplishments that we're proud of
We were able to have the agent's progress sync up with the frontend, making it very understandable for the user to see what the agent was actually doing when working with the database.
What we learned
Each of us learned a ton about the individual parts we worked on. John was responsible for frontend and connecting the model's actions to the frontend was a challenge, Liam learned a ton about tooling langchain to create a robust agent that could work with the fragmented database, Alex learned a ton about bringing the database together and prompt engineering.
What's next for Sonder Junior Analyst
We will be getting our product in front of as many small banks as possible, continuing iterating, and look to get a bank using our technology very soon.
Log in or sign up for Devpost to join the conversation.