Mitigram has launched an AI agent that turns “fragmented” bank risk pricing requests and communications into a single pool of data to inform trade finance decisions.
The company said the agent, named Alfred, will transform trade finance data that is “routinely buried in vast email chains” into a structured workflow that generates “actionable” risk pricing intelligence.
Banks can activate the AI agent simply by forwarding emails to it, and Alfred will “work quietly in the background, capturing and structuring pricing requests and related communications directly from existing channels, including email”, Mitigram said, with “minimal disruption” to existing workflows or legacy setups.
The agent has two main functions, including “parsing data from incoming emails to create a pricing request record in Mitigram’s system, and generating a draft email response that mirrors the language and tone of the original message”, the company’s CEO, Josh Kroeker, told GTR.
The resulting intelligence then flows into Mitigram’s established platform, allowing banks’ trade finance teams full visibility into their clients’ activity, as well as across the fintech’s large multibank corporate network to help them track opportunities, identify demand patterns and measure performance.
Currently, “all pricing requests [banks] get from clients who don’t want to use the multi-bank platform are sent to them by email, and all that data is lost”, Kroeker said.
“They don’t know the demand for certain markets, they don’t know the demand for certain issuing banks, and they don’t see what’s actually translating into a deal.
“All we do is allow that relationship managers, or whoever received the email, to forward it to our agent, and it all gets logged as data that allows for centralised responses. Now their data pool is their entire business.”
Aggregate market intelligence is pooled only when at least three corporates request a quote from an issuing bank, and at least three banks are quoting, limiting market-level signals to large, multi-banked transactions.
Single-bank or single-corporate interactions are “never included in aggregated data”, Kroeker said.
Individual bank data is instead held in a secure, segregated database, he added, and can be surfaced by the bank either via Mitigram’s Power BI tools within the platform, customised reports or as a raw data feed into the bank’s own systems.
Additionally, even when a transaction falls below the aggregation threshold, the broader market data “is still very useful for banks to quote, as it gives them an indication as to what the market is being priced for top-tier clients and they can use that as a ‘prime’ rate when quoting their own clients,” Kroeker said.
Alfred will begin by supporting export letter of credit confirmation and discounting requests, covering issuing bank risk, with Mitigram saying it plans to expand to additional trade finance instruments “over the coming weeks and months” in partnership with its banking clients.
Kroeker added the agent’s rollout came with “zero change for corporates and the relationship manager”.
“Alfred is going to help banks do what was previously impossible and make it possible,” he said.
“This is not just about taking your existing flows and moving them into a platform. It’s about reducing change, and I think this is the biggest advantage of these tools.”
