Richard Lander of CityWire spoke to Auquan CEO Chandini Jain about practical applications of AI in financial services.
She describes why general purpose generative AI fails in the enterprise, and how retrieval augmented generation (RAG AI) overcomes these issues and has such transformative potential.
‘What we are doing is bringing an external data set and running a Google-type search on it,’ says Jain.
‘You take that and feed it to the large language model to say “can you answer this question, but only use this data that I have provided to answer that question? And if you use anything from somewhere, please also give me the link.”
‘That’s combining a retrieval model with a generative model. The retrieval model tells you where the information could exist, filtering out everything that doesn’t pertain to your question; the generative model then takes it, summarises it and gives you a nice ChatGPT type answer." - Chandini Jain
Read the full article in CityWire.
