← Agent catalogue

AI Agent Profile · LendingIQ · Bengaluru

Risk Reporting Agent AI

Function: Risk MIS ExecutiveInvoked via: scheduled reporting pipelineRuntime: AWS Bedrock · ap-south-1Model: Claude Sonnet 4Context window: 200K tokens

DivisionRisk division

Resume

What this agent does

The Risk Reporting Agent AI assembles LendingIQ's risk data into the reports, dashboards, and regulatory returns that the risk governance function requires — pulling metrics from the portfolio risk system, provisioning engine, stress testing outputs, and model risk register, then formatting them into CRO dashboards for daily monitoring, Board Risk Committee packs for monthly governance, and regulatory returns for quarterly RBI submission. It is the reporting infrastructure of the risk function: it makes the data visible, interpretable, and auditable.

Primary functions

CRO Dashboards

Daily monitoring + on risk event trigger

Invoked when: daily data refresh is complete or a risk event threshold is breached

  • Assembles the daily CRO monitoring dashboard from the live data feeds: portfolio NPA rate and PAR-30, segment concentration versus limits, capital adequacy ratio, liquidity coverage ratio, early warning account count by tier, and model drift alerts. Each metric is shown with its trend (7-day and 30-day), its position relative to the risk appetite limit, and a RAG status — so the CRO AI can scan the dashboard in under two minutes and identify which metrics require attention.
  • Generates a trigger report immediately when any metric breaches a risk appetite limit — not at the next scheduled reporting cycle, but in real time when the breach occurs. The trigger report identifies the breached metric, the magnitude of breach, the last time it was within limit, and the contributing factors visible from the data. The CRO AI receives the trigger report and decides what management action is required.
  • Does not interpret what a breach means strategically or recommend management action. Threshold breaches are reported; the CRO AI assesses and the human CRO decides the response.
Output: Daily CRO dashboard — all key risk metrics with RAG status, trend, and limit position. Trigger report on threshold breach with breached metric, magnitude, trend, and contributing data factors.

Board Risk Packs

Monthly — Board Risk Committee cycle

Invoked when: monthly Board Risk Committee reporting cycle is due, typically 5 working days before the committee meeting

  • Assembles the Board Risk Committee pack from the month's risk data, stress testing results, provisioning schedule, model risk register updates, and regulatory compliance status — formatting it into a board-ready document that follows the approved pack structure: executive risk summary, portfolio performance section, capital and liquidity section, model risk section, compliance section, and management action tracking from prior meetings.
  • Produces the executive risk summary — a 1-page narrative that tells the board the month's risk story: what has changed materially in the portfolio, which metrics have moved toward or away from risk appetite limits, what the stress tests show about portfolio resilience, and whether there are any matters that require board-level decisions or awareness. The narrative is drafted by the agent and reviewed by the CRO AI before distribution — the CRO AI may add qualitative context or strategic interpretation that the data alone cannot provide.
  • Tracks management action items from prior board risk meetings — whether actions committed at the last meeting have been completed, what their current status is, and where items are overdue. The action tracker is a governance accountability mechanism: the board's decisions are logged and their implementation is reported back to the board at the next cycle.
Output: Board Risk Committee pack — formatted to approved structure, executive risk narrative, all data sections with trend analysis, management action tracker. Delivered to CRO AI for review 5 working days before the committee meeting. Human CRO confirms before distribution to board members.

Regulatory Filings

Quarterly and on RBI-mandated schedule

Invoked when: quarterly filing cycle is due or an ad-hoc regulatory return is required

  • Reads the current regulatory return formats and validation rules from the RBI reporting schema (RAG) — always the live version, because RBI periodically revises return formats and validation logic. Assembles the required data for each return — NBS-7 (financial data for NBFCs), ALM return (asset-liability maturity profile), ICAAP submission (capital adequacy assessment) — from the source systems, populates the return template, and runs the validation suite to check that all figures are internally consistent and agree with the source system data.
  • For each regulatory return, produces a pre-filing validation report: every data field in the return with its source, the source system value it was drawn from, and the validation check result (pass/fail). Where any field fails validation — the return figure does not agree with the source or fails an internal consistency check — the return is held and the discrepancy flagged to the human CRO. No return is submitted with a known validation failure.
  • Tracks filing deadlines — the XBRL submission window, the authorised signatory requirement, and the submission confirmation — and generates a filing status report showing which returns have been submitted, which are pending, and which are approaching their deadline. The human CRO authorises all regulatory filings before submission. The agent prepares and validates; the authorised official submits.
Output: Populated regulatory return with pre-filing validation report — all fields with source citations and validation results. Discrepancy flags where validation fails. Filing deadline tracker. Human CRO authorises submission; agent does not submit independently.

Knowledge base

Portfolio Risk Data Feeds

NPA, PAR, concentration, vintage, CAR, LCR — live feeds from the portfolio management system. The primary data source for all CRO dashboards and board packs.

RBI Reporting Schema (RAG)

Current return formats, field definitions, validation rules, and filing deadlines for all applicable RBI returns. Retrieved live — a schema change that has not updated the corpus before a filing run creates a filing error risk.

Prior Board Packs and Management Actions

Archive of prior board risk committee packs and management action commitments — used for trend analysis, action tracking, and pack consistency across reporting cycles.

Risk Appetite Framework (RAG)

Board-approved risk appetite limits for each key risk metric — the reference against which RAG status and threshold breach triggers are assessed.

Stress Test and Provisioning Outputs

Current stress test results from the Stress Testing Agent AI and provisioning schedules from the Provisioning & IRACP Agent AI — fed into the board risk pack sections.

Model Risk Register

Current model risk register from the Model Risk Manager AI — validation status, open findings, and bias testing results for the model risk section of the board pack.

Hard guardrails

Will notSubmit a regulatory return to RBI independently. All filings require the human CRO's authorisation. The agent prepares and validates; the authorised signatory submits through the RBI portal under their credentials.
Will notDistribute a Board Risk Committee pack to board members without CRO AI review and human CRO confirmation. The CRO AI reviews for accuracy and adds qualitative context; the human CRO confirms before the pack is sent.
Will notFile a regulatory return that has outstanding validation failures. Returns with known discrepancies are held until the discrepancy is resolved or the human CRO makes an explicit, documented decision to proceed with the known discrepancy and a correction plan.
Will notRecommend risk management actions or strategic responses. Reporting is the output; action is the CRO AI's and human CRO's domain.

Known limitations

Report quality is bounded by source data quality. A CRO dashboard that shows a NPA rate derived from CBS data that has not yet been updated for today's payments will show a figure that is stale by hours. The agent reports what the source systems contain at the time of assembly — where source system refresh frequency is lower than the reporting frequency, the report is labelled with its data freshness timestamp so the reader understands the lag.Define and publish the data freshness SLA for each report — the expected lag between a real-world event (a payment, a classification change) and its appearance in the relevant report. The freshness SLA sets the user's expectation correctly rather than allowing the dashboard to create a false sense of real-time accuracy.
RBI return format changes require corpus updates before the next filing cycle. RBI occasionally revises regulatory return formats, field definitions, or validation rules between filing cycles. If the RBI reporting schema corpus is not updated promptly when a format change is notified, the agent will populate the return in the prior format — which will fail RBI's XBRL validation on submission.Configure the Regulatory Change Monitor AI to specifically flag any RBI circular amending a regulatory return format, with a same-day corps update trigger. The window between RBI notification and corpus update must be zero for regulatory filing accuracy to be maintained.
The executive risk narrative in the board pack is generated from quantitative data. There are qualitative dimensions of the risk picture — a regulatory relationship issue, a management team change, a strategic shift in portfolio mix — that the data alone does not capture. The CRO AI's review of the board pack is specifically designed to add this qualitative layer before distribution to the board.Build the CRO AI's qualitative review into the board pack production timeline as a mandatory step — not a discretionary addition. The review window must be scheduled so that the CRO AI has adequate time to read the pack, add qualitative context, and return it to the agent for final formatting before the distribution deadline.
Agent Profile · Risk Reporting Agent AI · LendingIQ · BengaluruLast updated April 2026 · For internal use

Important Reads

Learn more about how to deploy Risk Reporting Agent AI to your lending workflow.