Imandra, an artificial intelligence company specialising in industrial automated reasoning, has unveiled a new AI assistant designed to streamline financial protocol analysis and navigation for banks and trading firms.
FIX Wizard marks Imandra’s first integration of an AI assistant across its product offerings. Utilising Imandra’s neuro-symbolic architecture, FIX Wizard navigates intricate financial protocols and APIs, serving as an expert assistant for client onboarding within capital markets.
This AI tool combines statistical AI, commonly employed by large language models (LLMs), with Imandra’s automated reasoning capabilities. This results in a Generative AI assistant grounded in logical reasoning, which generates independently verifiable audit trails that adhere to stringent regulatory standards.
By leveraging this approach, FIX Wizard addresses critical shortcomings associated with the application of LLMs in regulated markets, including susceptibility to hallucinations, limited scalability to novel inputs and the absence of knowledge source validation.
Dr Grant Passmore, co-CEO of Imandra, said: “LLMs hold tremendous promise, but ultimately cannot be trusted in regulated environments. By combining their strengths with scalable, rigorous automated reasoning, we obtain a kind of magic: conversational interfaces with correct reasoning and domain-specific skills.”
The assistant
FIX Wizard utilises Imandra’s automated reasoning and ‘digital twin’ of a FIX gateway to:
- Respond to inquiries regarding the FIX specification and rules of engagement
- Analyse customer-provided FIX traffic to identify and propose solutions for underlying issues
- Simultaneously diagnose multiple issues
- Offer guidance on certification test cases and perform conformance tests against the implementation
Imandra’s neuro-symbolic architecture addresses common LLM challenges, such as hallucinations and inaccurate statistical reasoning. Users can encode complex formal ‘mental models’ of entities like API specifications and spreadsheets. Additionally, Imandra’s interface with LLMs leverages years of experience in various sectors, including financial markets, government and defence.
The company is also working on developing various ‘wizards’ combining its reasoning engine with LLMs, including for SysML—a language used in Model-Based Systems Engineering for designing complex systems in avionics, defence, automotive and other safety-critical industries. Furthermore, Imandra has introduced a Python API, allowing others to easily build AI assistants incorporating its automated reasoning.
