AI Data Retro

• 

  Interaction

• 

  Data complexity

• 

  Arctitecture

Inspiration

Many central operations teams in banks still rely on manual reconciliation across systems, slowing operations, audit challenges, and raises risk as data volumes grow.

What it does

An AI-powered agent solution that understands natural language, orchestrates multi-system validation, and delivers actionable insights without manual intervention

How we built it

Develop an AI assistant using NLP, sentence embeddings, and intent detection to interpret commands, confirm intent, and automate system reconciliation.

Challenges we ran into

It took me something to create a logic to pull the data from the report excel sheet

What we learned

It is now easier to compare the data between the reports and back end

What's next for AI Data Retro

Planning to upscale by integrating multiple excels and multiple databases

Built With

• all-mini-lm-l6-v2
• mysql
• openpyxl
• padas
• python
• sqlite
• streamlit