Mon, Dec 23 2024
A new AI assistant from Imandra, an artificial intelligence startup that specializes in industrial automated reasoning, is intended to make financial protocol analysis and navigation for banks and trading companies easier.
Imandra is introducing an AI helper into its product line for the first time with FIX Wizard. FIX Wizard is an expert helper for client onboarding in capital markets, navigating complex financial protocols and APIs with the help of Imandra's neuro-symbolic architecture.
This artificial intelligence application integrates Imandra's automated reasoning skills with statistical AI, which is frequently used by large language models (LLMs). This leads to the creation of a logically based Generative AI assistant that produces independently verified audit trails compliant with strict regulatory requirements.
FIX Wizard utilizes this method to tackle significant issues related to the use of LLMs in regulated markets, such as their vulnerability to hallucinations, their inability to scale to new inputs, and their lack of knowledge source validation.
"LLMs hold tremendous promise, but ultimately cannot be trusted in regulated environments," stated Dr. Grant Passmore, co-CEO of Imandra. We are able to create conversational interfaces with accurate reasoning and domain-specific abilities by fusing their advantages with scalable, rigorous automated reasoning.
The helper
FIX Wizard makes use of the "digital twin" of a FIX gateway and Imandra's automated reasoning to:
Answer questions concerning the rules of engagement and the FIX specification.
Examine FIX traffic supplied by the client to find and suggest fixes for any underlying problems.
Diagnose many problems at once
Provide direction on certification test cases and conduct testing of conformance with the implementation
Hallucinations and erroneous statistical reasoning are two typical LLM problems that are addressed by Imandra's neuro-symbolic architecture. Complex formal "mental models" of entities such as spreadsheets and API specifications can be encoded by users. Imandra's interface with LLMs also benefits from years of experience in a variety of industries, including as the financial markets, the government, and the military.
A variety of "wizards" that combine the business's reasoning engine with LLMs are being developed by the company. One such wizard is for SysML, a language used in Model-Based Systems Engineering to design complex systems for the avionics, defense, automotive, and other safety-critical industries. Moreover, Imandra has released a Python API, making it simple for third parties to create AI assistants that use its automated reasoning.
Leave a Comment