When a vendor has an incident, know what breaks in your bank before the reporting clock runs out.
Fortivance connects vendor incidents, contracts and audit reports to the banking services that depend on them. It shows a third-party risk team which customer services are exposed, how much of the six-hour RBI reporting window is left, and whether the bank holds the contractual evidence regulators expect. AI does the reading and drafting; people make every decision.
Why this matters
Banks increasingly run on cloud providers, payment processors and technology vendors, and those vendors run on their own subcontractors. When one link fails, the bank is still accountable to its customers and its regulator.
RBI's outsourcing Directions for commercial banks (28 November 2025) also require a central inventory of outsourced IT services, mapping of dependencies on third parties and their key subcontractors, and contracts that give the bank and RBI audit rights and timely incident notification.
The gap Fortivance closes
Most vendor risk tools answer "is this vendor risky?". During an incident, the questions a bank actually has to answer are different.
| Question | Typical approach today | With Fortivance |
|---|---|---|
| Which of our services are affected? | Calls and spreadsheets across IT, operations and vendor management while the clock runs. | Impact traced in seconds from the vendor, through its subcontractors and the bank's applications, to each customer-facing service and its impact tolerance. |
| How much time is left to report? | Worked out by hand, often from the time the notice arrived rather than when the vendor detected the incident. | A live countdown from the vendor's stated detection time, with the vendor's notification delay shown alongside it. |
| Is this notice genuine? | Rarely checked under time pressure. | Sender domain checked against the vendor registry; embedded instructions aimed at automated systems are flagged and ignored. |
| Do our contracts actually cover this? | A certificate or SOC 2 report on file is treated as proof, even when it does not cover the contract terms or carves out a subcontractor. | Each requirement mapped to the exact clause that satisfies it, or flagged as missing, with the source text shown for a reviewer to verify. |
What Fortivance does
Incident intelligence
Upload a vendor notice. Fortivance extracts the facts, identifies the vendor and any subcontractor involved, computes the impact, starts the reporting clock and drafts a response plan for approval.
Open incidentsDependency intelligence
A live map of vendors, fourth parties, applications and business services, with single points of failure and fallback options marked.
Open dependency graphCompliance intelligence
Contracts and assurance reports split into citable clauses and mapped to RBI outsourcing requirements. Analysts verify or dispute each finding; confirmed gaps become actions.
Open evidenceVendor intelligence
Vendor profiles with regulatory scope, overdue reviews, possible mis-tiering, and a rule-based risk score where every point is explained.
Open dashboardContinuous evaluation
Extraction accuracy, evidence status agreement and citation rejection rates measured against hand-labelled gold data.
Open evaluationHow an incident moves through Fortivance
AI is used only where language has to be read or written. Everything that must be exact is ordinary, testable code.
Measured in this workspace
These figures are calculated live from the incidents and evidence in this instance, not estimated.
How Fortivance uses AI responsibly
- People decideEvery AI output is marked as needing review. Response plans need an approver; evidence findings need an analyst to verify or dispute them.
- No regulatory verdictsFortivance shows the reporting window but never decides whether an incident is reportable and never files a report.
- Citations that can be checkedThe AI may cite only clause identifiers from the document. Fortivance checks that each one exists and shows the original text; anything it cannot verify is sent back for review.
- Untrusted input stays untrustedUploaded notices are treated as data. Instructions hidden inside them are flagged and ignored, and a notice from an unrecognised domain gets no automatic plan.
- Explainable risk scoresVendor risk scores are rule-based. Each contributing factor and its points are listed.
- Controlled costResults are cached, every AI call is logged with its token usage, and a daily spending cap is enforced.
- Measured against gold dataAI outputs for incident extraction, evidence mapping and citations are continuously measured against hand-labelled data on the Evaluation page.
Who it is for
About this instance
This is a demonstration built on synthetic data. Deccan Meridian Bank Ltd and every vendor, contract, report and incident shown are fictional. Regulatory paragraph references marked "(verify)" should be checked against the official RBI text before being relied on. Fortivance supports analysts; it is not legal or regulatory advice.