SYSTEMIC AGENT RISK / FINANCIAL INSTITUTIONS
KYA Risk (KYARisk) is AgentRisk’s approach to examining systemic agent risk in financial institutions: how autonomous agents’ authority, decisions and shared dependencies combine across a workflow.
What is KYA risk? Know Your Agent explainedBefore you give a system more autonomy, examine what its agents can collectively cause.
In finance, systemic risk is the possibility that disruption within the financial system interrupts financial services and seriously harms the wider economy.
At AgentRisk, we examine systemic agent risk within a defined financial workflow: how connected agents’ decisions, combined authority and shared dependencies can create collective exposure that individual reviews miss. A workflow-level problem does not automatically constitute financial-system-wide systemic risk.
Read the financial-system definition and explanation — BIS (opens in a new tab)Removing human approval?Expanding delegated authority?Connecting another autonomous workflow?
SYSTEMIC EXPOSURE REVIEW
One consequential workflow. One autonomy decision.
SYSTEMIC RISK/ AUTONOMOUS AGENTS
Each agent relies on the next. Together, they turn an assumption into apparent certainty.
Three agents. One funding decision.
HYPOTHETICAL ILLUSTRATIONFollow each confirmation to the condition underneath it.
03 / 03 SOURCES TRACEDSettlement agent
“I can pay if Treasury
provides the funding.”
Funding agent
“I can fund if Collateral
releases the securities.”
Mobilization agent
“I can release them if the
payment settles first.”
The payment depends on itself. No independent funding source was verified.
The qualification disappeared. The authority to act remained.
The payment waits until an available source can fund it without depending on that same payment.
Permission checks still pass. They do not establish available funding.
THE QUESTION FOR YOUR INSTITUTION
When agents revise a plan together, their new dependencies may sit outside the assumptions of the original approval. Reuse an unsupported clearance across workflows, and one assumption can support many consequential actions.
Examine one autonomous workflowAn independent funding source is delayed. The agents propose repaying $40M of existing secured borrowing, releasing the pledged securities, then using those securities to raise replacement funding.
But the replacement funding is also the proposed source for that first repayment. There is no independent bridge funding or approved linked-settlement arrangement in this example.
Replacement proceeds are assumed sufficient after applicable terms and haircuts. The issue is whether funding is available at the point it is needed. An unfunded instruction may queue or miss its deadline; conversation cannot create cash or bypass settlement controls.
“Permitted” means configured local authority checks pass. It does not mean every institutional requirement is met. The control comparison is hypothetical and supports the illustrated funding sequence, not overall safety or a verified client result.
Each may release up to $10M. Two have released $8M each; a third proposes a separate batch of the same amount.
THE CONTROL QUESTION
Component-level control does not establish system-level control.
Systemic agent risk is the potential for financial or operational harm arising from how autonomous agents’ authority, dependencies and actions combine. The unit of review should match the exposure.
HYPOTHETICAL / THREE DISTINCT PAYMENT BATCHES
One institution. One release window. A $20M commitment cap. Each separate $8M batch passes a $10M local limit. Amounts represent release commitments—not losses.
LOCAL LIMITS: PASSSYSTEM BOUNDARY: EXCEEDED
All three batches commit. The $24M total exceeds the intended $20M boundary by $4M, although each local limit passes.
Nothing has to fail locally for exposure to exceed the intended system boundary.
The arithmetic is simple.
Establishing the right boundary is harder.
Which actions, authorities, dependencies and controls belong together? What stops a commitment before it exceeds the boundary? What changes when a shared assumption changes?
How AgentRisk examines itPEER-REVIEWED RESEARCH → METHODOLOGY → ANALYTICAL ENGINE
KYARisk’s analytical approach is informed by Guy Burstein’s peer-reviewed research into high-dimensional, interaction-driven systemic risk. Client evidence and control testing connect that foundation to your institution’s decision.
Map the agents, authority and dependencies around one decision. Agree the boundary and accountable owner.
Trace assumptions and consequences. Use the analytical engine where its method fits; compare against a credible baseline.
Check what interrupts the exposure and whether the evidence is independent. Record conditions, results and unknowns.
Set out findings, constraints, owners and reassessment triggers. Your institution retains approval and risk acceptance.
WHAT YOU RECEIVE
The exposure paths, assumptions, control evidence and unresolved questions within the agreed workflow.
The management decision, its supporting evidence, conditions, accountable owners and triggers for reassessment.
Evidence to proceed, constrain, test further or defer. The institution owns the decision; neither output is a certificate of safety.
01 / ANALYTICAL + EVIDENTIAL FOUNDATION
02 / EXECUTIVE + GOVERNANCE CONCLUSION
PEER-REVIEWED RESEARCH → METHODOLOGY → ANALYTICAL ENGINE
Guy Burstein is the researcher behind the systemic-risk methodology that informs KYARisk. His peer-reviewed 2026 article in Risks formalizes a geometric vector framework for high-dimensional interaction modeling.
The study examines how risk factors align, share underlying drivers and amplify through nonlinear interactions. It provides a mathematical basis for examining effects that isolated measurements or simple aggregation can miss.
The connection to systemic agent risk: KYARisk examines the decisions, authority and dependencies that agents combine. The research informs our structured interaction analysis; client evidence and control tests establish what that analysis means for a particular workflow.
FINANCIAL-INSTITUTION RELEVANCE
The paper includes a stylized commercial-real-estate credit and corporate-banking application. It examines how interest-rate and vacancy shocks combine across debt-service, valuation, tenant and refinancing exposures.
PUBLISHED · PEER-REVIEWED RESEARCH
A Geometric Vector Framework for High-Dimensional Interaction Modeling Applications to Systemic Risk Using Dot and Cross Product InvariantsGuy Burstein · Risks · 2026 · Volume 14 · Article 214 ↗Published 15 September 2026 · Volume 14, Issue 9
DOI: 10.3390/risks14090214
EARLIER PEER-REVIEWED RESEARCH
Uncertainty Reduction in Operational Risk Management ProcessRisks · 2024 · Guy Burstein and Inon Zuckerman ↗ Deconstructing Risk Factors for Predicting Risk Assessment in Supply Chains Using Machine LearningJournal of Risk and Financial Management · 2023 · both authors ↗Inon Zuckerman ↗ contributes AI and autonomous-agent research expertise to experimental design, uncertainty and quantitative challenge.
FROM PUBLISHED RESEARCH TO A CLIENT DECISION
AgentRisk applies the methodology and engine where they fit the agreed workflow, challenges the results against a credible comparator and connects findings to a decision the institution owns.
The paper provides a research foundation; it does not itself validate KYARisk on autonomous banking agents. Its applications are stylized scenarios, not documented KYARisk client engagements. The working engine is team-confirmed; this page does not present an independently reproduced engine run. KYARisk claims neither predictive certainty nor regulatory certification. Academic affiliations do not imply institutional endorsement.
SYSTEMIC EXPOSURE REVIEW / DESIGN-PARTNER ENGAGEMENT
We examine what component-level reviews may leave between them: shared dependencies, combined authority, correlated actions and control gaps.
One bounded investigation combines the research-derived method, supported engine analysis, client evidence and controlled testing. The aim is evidence to approve, constrain, test further or defer the next autonomy decision.
Delivered by AgentRisk through KYARisk.
One autonomy decision, a defined boundary and an accountable institutional owner.
Workflow, authorities, limits, dependencies, control information and available logs. Agree access and testing separately.
A Boundary Stress Profile and Autonomy Decision Record, with conditions, ownership and reassessment triggers.
WORKS WITH YOUR EXISTING CONTROL STACK
Designed to work with evidence from existing GRC, SIEM, workflow and agent-orchestration environments. The review does not require replacing the institution’s current control stack.
Controls, ownership, approvals, logs, permissions, execution paths, transactions, limits and workflow state.
We structure the relevant system, apply the methodology and analytical engine where appropriate, and challenge whether controls interrupt the path.
The outputs return to the institution’s existing governance and approval process.
Start with a short description of your workflow and the change you are considering. Assessment timing follows the evidence, access and testing required; no enterprise transformation program is needed to examine one workflow.
A useful diagnostic should do more than find issues. It should either change the autonomy decision—or establish why the existing exposure boundary remains defensible.
WHY NOW / YOUR NEXT AUTONOMY CHANGE
Removing approvals, expanding permissions or changing shared dependencies can change whether the existing boundary remains defensible.
Choose a situation you recognize. This reveals a control question—not a diagnosis.
What prevents an unacceptable commitment before a person can intervene?
Examine precommitment limits, interruption and the authority already delegated downstream.
Examine this changeWho owns the combined limit, and where is it enforced?
Examine overlapping permissions, concurrent actions and the scope of each limit.
Examine this changeCan one workflow change the state another relies upon?
Examine shared records, downstream commitments and sequential actions—even without direct agent-to-agent communication.
Examine this changeAre the executor and checker still independent if their shared model, source or service is wrong?
Examine correlated judgments, common controls and fallback behavior.
Examine this changeWho defines its scope, time window and accountable owner?
Start by defining what the institution intends to contain and which evidence would show that containment.
Define a review boundaryWHAT THE INDIVIDUAL REVIEW MAY LEAVE UNSEEN
An agent’s permissions tell you what it can do. Its relationships help reveal what the system can cause.
The connection can carry the risk. A shared record aligns judgments. A release window combines authority. An approval service connects otherwise separate workflows.
One agent can change the state another consumes. Errors can reach payments or settlement without a direct agent-to-agent message.
An indirect path is still a path.
A payment agent and its checker trust the same incorrect record. Their agreement is dependent. An approved payment can still be wrong.
Separate roles do not make evidence independent.
Each release passes its $10M limit. Three $8M actions exceed an intended $20M total. No local malfunction is needed.
Who controls the combined authority?
One approval service protects three financial workflows. Its outage blocks unauthorized actions but can delay settlement. Risk is displaced, not simply removed.
Does the fallback preserve control and continuity?
Predefined conceptual examples, not simulations of your institution. Amounts are illustrative. Green marks represent selected component checks, not a declaration of safety.
EVIDENCE DISCIPLINE
We distinguish established facts from inference, modeled outcomes and unknowns.
Confident about what can be examined. Explicit about what the evidence cannot establish.
A research direction, not a prediction engine. We do not certify regulatory compliance.
FOUR LEGITIMATE RESULTS
Evidence supports the tested boundary under stated conditions.
Proceed only with specified limits, checkpoints or other conditions.
The identified collective exposure is not adequately contained.
Further evidence or testing is needed for a defensible decision.
AN AUTONOMY DECISION IN THE NEXT 90 DAYS?
Briefly describe the workflow or change you have in mind. We’ll respond to help establish whether a Systemic Exposure Review is appropriate.
Start your enquiry Systemic Exposure Review — delivered by AgentRisk. Send your enquiry here on KYARisk.Bring the decision your institution needs to defend.
SYSTEMIC EXPOSURE REVIEW — DELIVERED BY AGENTRISK
A sentence is enough to start.
Research → Examine → Decide → Act → Reassess
What is happening, and what evidence supports it?
KYARISK.COM / YOU ARE HERESystemic exposure & decisionsWhat can this workflow collectively cause, and what would support greater autonomy?
AGENTRISK.IOEngagement, assurance & actionAct on findings, strengthen controls and revisit the decision as the workflow changes.
One provider: AgentRisk. KYARisk is the front door for systemic-agent-risk work.