Imandra has today launched the firm’s first artificial intelligence (AI) assistant for its FIX Connectivity suite used by banks and trading firms.
Named FIX Wizard and built on Imandra’s neuro-symbolic architecture, the new offering automatically reasons complex financial protocols and APIs, serving as an AI assistant for onboarding clients in capital markets.
Statistical AI used by LLMs and Imandra’s automated reasoning are combined by FIX Wizard to create a generative AI assistant rooted in logic, with verifiable audit trails that meet regulatory requirements.
Imandra stated that this approach addresses the fundamental shortcomings of applying LLMs in regulated markets, which include hallucinations, lack of scalability to unseen inputs and lack of validation of knowledge sources.
“The problems with LLMs are well documented these days – ‘hallucinations’, a lack of correctness and transparency. LLMs use statistical methods to generate answers, essentially by predicting the next word in a sequence. LLMs are exceptionally useful in language translation, but they are incapable of reliable reasoning, i.e. thinking logically about a problem and deriving correct results with audit trails,” Denis Ignatovich, co-founder and co-CEO at Imandra, told The TRADE.
“Imandra’s automated reasoning engine uses rigorous logical inference (“symbolic AI”) to derive results, and it is already relied upon by leading firms in financial markets and safety-critical industries – in fact, our Imandra Markets product already successfully supervises exchange systems responsible for around 25% of all European equities traded volume.”
The new offering uses Imandra’s ‘digital twin’ of a FIX gateway and automated reasoning to answer questions about the FIX specification and rules of engagement, and analyse customer-provided FIX traffic, understand underlying causes of issues and make recommendations on how to address these.
FIX Wizard is also able to diagnose various issues at once as well as provide guidance on certification test cases and run conformance tests against the implementation.
“With our neuro-symbolic architecture, we combine the linguistic strengths of LLMs with the logical capabilities of our reasoning engine, resulting in a fluid conversational interface to an automated expert with domain-specific reasoning skills,” Grant Passmore, co-founder and co-CEO at Imandra, told The TRADE.
“In this case, we infuse precise logical specifications with the LLM, confining it to the specific FIX requirements, making our FIX Wizard a reliable and compliant AI assistant that not only processes language but, crucially, understands it within the context of the FIX protocols.”