Every signal is read as it lands. The Credit Risk officer opens the inbox to ranked exposures, options for action and the reasoning behind them.
Animated demo: scattered credit signals (news headlines, exposure tickers, margin calls) converge into a tiered counterparty treemap. An officer then accepts a tightening recommendation on BP, the largest exposure, which recolours from immediate to elevated.
The mid-tier counterparty that didn't make this morning's priority list. The exposure looked fine yesterday. New trades came in. Payments shifted. The market moved. News broke at eleven. By the time it surfaced in the afternoon review, two other desks had already traded on it. The signal wasn't missing. Nobody had the time to connect it.
Ranked, with the reasoning written up overnight. No morning report to read through; the agent has already done it. The hour-per-call becomes minutes.
The exposure reviewed, the agreements checked, the dispute history pulled, the precedent verified, the wording drafted. An hour for one decision. During a stress event, the firm that acts first secures protection while it is still available.
A counterparty breaches a threshold. By the time the officer opens the alert, the Exposure Explanation Agent has decomposed the move: which trades drove it, which limits were impacted, what offsets exist and what actions are available. Three ranked options appear on screen with the supporting rationale attached.
Credit committee asks why the limit was raised on a name now in difficulty. The reasoning lived in someone's head. The email trail has the conclusion but not the workings. Reconstructing the call is costly and incomplete.
By the close, every decision made today has its reasoning recorded. The committee in six months opens the case and the workings are already there.
Existing credit platforms hold the limits, the exposures, the collateral and the counterparty master. They are the system of record. The work between the data and the decision is still hours of officer time. Surity is the intelligence layer that closes that gap. The system of intelligence.
Traditional credit platforms hold the limits, exposures, collateral and counterparty master. Excellent at storing the state of the portfolio and presenting it on a screen.
None of the real work is software the system does today. It's hours of officer time: watching every counterparty every day, reading the news, aggregating across legal entities and tenors, spotting the trigger, then writing it up so a credit committee can stand behind it.
A faster query interface doesn't change the shape of the work. It just compresses one step.
Surity sits above the system of record. Eighteen specialist agents monitor limit utilisation, exposure changes, margin events and counterparty news, around the clock, on every name in the portfolio.
Each trigger lands as a typed issue, tiered immediate / urgent / elevated / routine, and routed to the agent best equipped to reason about it. Every agent has a fixed vocabulary of recommended actions.
Every agent run, tool call, recommended option and officer response - confirm, override or dismiss - is persisted in the decision log. That is the difference between a clever assistant and a system the credit committee can stand behind six months later.
Eighteen agents, each scoped to a single credit problem: limit utilisation, exposure explanation, margin-call lifecycle, collateral coverage, counterparty news, exposure anomaly, deal readiness, netting review, payment review, portfolio aggregation, signal synthesis, and the Credit Overseer that holds the chat thread. Each has its own tools, escalation rules and a fixed recommendation vocabulary. The LLM orchestrates. The credit logic is explicit and auditable.
Eleven triggers live today: limit utilisation, margin-call lifecycle, collateral coverage and expiry, EFET/ISDA/CSA documentation, exposure change and anomaly, counterparty news, netting review, gross/net drift, payment review, early warning, signal synthesis. A further six (covenants, ratings, settlement lifecycle and others) are in build. Credit doctrine, not technology, decides what's next.
Every agent recommendation comes back as a ranked set of options with an explicit recommendation, validated against a schema. The officer confirms, overrides, or dismisses. There is no "the model said". There is a record of what the model proposed and what the officer decided.
Every run, every tool call, every reasoning step, every officer response, captured and explainable. Designed from day one to be defensible to credit committee, internal audit, and external regulator.
“A faster dashboard doesn't change the shape of credit work. An intelligence layer that watches every counterparty every day, with the reasoning recorded and the officer still in the loop, does.”David Priestley, founder
Career entrepreneur across finance and commodities technology. Founder of DPA, Raft International (AIM-listed risk and trading), CPQI, MDX Technology and CubeLogic. Two-time EnergyRisk award winner. Founder and funding partner behind Surity.
LinkedIn
Thirty years building front-office technology for capital markets, specialising in high-volume, low-latency data and quantitative analytics. Twelve years at Credit Suisse leading strategic risk platforms in derivatives. Founded Schematiq and led it to acquisition by MDX Technology, where he served as CPO.
LinkedIn
Successful entrepreneur with extensive experience founding, scaling, and exiting technology businesses across finance, climate, commodities and energy. Co-founder of Speedwell Climate (acquired by Vaisala), director at Gissing Software (acquired by Reuters), and co-founder and CEO of MDX Technology. Leads commercial strategy at Surity.
LinkedIn
Head of Credit for EMEA and LatAm at Mercuria, with global responsibility for Metals. Over 15 years of counterparty risk and credit governance inside one of the world's largest energy and commodities trading houses, covering crude, refined products, gas and power, LNG, environmental products and metals.
LinkedInTell us about your credit function. We respond within a working day.