In 2025, Visa processed a staggering 106 million disputes, a 35% increase since 2019. For card issuers, this surge has significantly increased the workload and amplified a long-standing operational bottleneck: the manual review.
With the Visa Global Technical Release 26.1 (late April 2026), two new AI-driven tools are becoming central to the issuer’s toolkit: Visa Dispute Document Analyser (VDDA) and Visa Dispute Intelligence (VDI).
If you handle disputes on the issuing side, understanding what these tools deliver and how to use them well has become a practical matter.
Inside the dispute review workload
When a cardholder disputes a transaction, the issuer files a chargeback, and the acquirer has a fixed window to respond with evidence that the transaction was valid. The dispute response is meant to be a structured evidence package mapped to the issuer's reason code. In reality, the volume and variety of supporting documents, from email threads to delivery confirmations to authentication logs, means case handlers still spend significant time extracting the data points that matter.
Depending on the case, this can take anywhere from ten minutes to half an hour. Multiply that across a dispute team's daily volume, and it becomes one of the most expensive parts of the operation.
Acquirers face the same workflow on the other side of the case. When they receive evidence from merchants to build the dispute response, and later when responding to issuer pre-arbitrations, they get the same mix of email threads, screenshots and logs, and have to transcribe the relevant data into the structured response format.
What the Visa Dispute Document Analyser (VDDA) is
VDDA is a service that uses AI and Optical Character Recognition (OCR) to scan attached dispute documents and produce a structured summary of critical data points across all dispute categories (fraud, authorisation, processing errors, and consumer disputes).
It is particularly useful in cases involving heavy evidence packs, such as first-party misuse (often called “friendly fraud”), where compelling evidence is buried in long authentication logs, email threads, or shipping records. While acquirers have had access to VDDA since late 2025, it is now becoming a core capability for issuers.
How VDDA turns an evidence pack into a structured case view
When a case includes attachments, a "Summarise with Doc Analyser" button appears. Once triggered, the AI takes up to a minute to generate a structured report with citations, allowing agents to verify the findings immediately.
The summary is divided into key sections that map to the dispute category:
- Overview – a high-level narrative of the document's contents.
- Purchase Transaction Detail – merchant info, IP address, device ID, transaction dates, shipping details (for Fraud and Consumer Dispute cases), amounts
- Order Detail – customer info, order lines, delivery tracking, flight, hotel, and vehicle rental records
- Service Detail – for Authorisation cases: tickets, flight and accommodation records
- Transaction Receipt – for Processing Error cases: receipt numbers, transaction data
- Citations – every extracted field references its source document, page number, and attachment ID, allowing agents to verify the AI's output instantly.
What VDDA delivers, and where human judgement still sits
While VDDA is a massive leap forward for efficiency, it has specific limitations that issuers must understand:
- Data extraction, not reason-code reasoning: VDDA is scoped to surface the facts in the documents. Mapping those facts to reason code logic (10.4 vs 13.9, for example) stays with the handler.
- Refund detection, not remedy assessment: VDDA flags that a refund is mentioned. Whether it meets the scheme's criteria for closing the case is the handler's call.
- Dense technical logs still need a human read: For cases like non-receipt of cash at an ATM, VDDA surfaces the extracted fields. Interpreting them against the cardholder's specific claim is still the handler's job.
Consider a cash withdrawal dispute, the non-receipt of cash. The acquirer typically responds with ATM logs and transaction records. VDDA pulls the decisive fields from those dense records in seconds, and the judgement on whether they remedy the claim under reason code 13.9 sits with the case handler, as intended.
How issuers can configure and use VDDA
With the April 2026 release, VDDA becomes available to all issuers as an opt-in feature. Each summary request carries a per-request fee. There is no fee for simply having VDDA enabled, only for actually pulling a summary.
Two things follow from this:
- Enablement is a conscious choice: Issuers explicitly opt in to make VDDA available on their cases, via the Visa Online (VOL) configuration portal or through the issuer's processor, and can switch it off again at any time.
- The economics are per-case: VDDA's value compounds with complex cases (long evidence packs, fraud disputes with multi-source authentication, multi-document pre-arbs). For straightforward cases, agents may choose to review directly.
What the Visa Dispute Intelligence (VDI) is
VDI is a predictive AI tool designed to support case-by-case analysis. By leveraging Visa’s global transaction data, it provides issuers with a "VDI score" that helps assess the legitimacy of a dispute.
VDI is the complementary piece. Where VDDA processes documents attached to a dispute response, VDI provides a stage-specific win-likelihood / outcome-support score, with key contributing factors, to help issuers and acquirers decide whether to submit, accept, decline, respond, or investigate further.
Like VDDA, VDI is request-based: the issuer explicitly triggers the calculation when they want the score, rather than having it run automatically on every case. Issuers can switch it on or adjust the setting at any time through their processor or the VOL configuration portal.
Strategic Alignment: VDDA & VDI within Amiko
At Rivero, we believe human dispute agents shouldn't have to jump between systems to access these efficiencies. That is why the Visa Dispute Intelligence (VDI) and the new Visa Dispute Document Analyser (VDDA) are fully supported in Amiko today, available directly within the case view.
We have ensured that our users have access to the Visa’s dispute intelligence stack:
- VDDA on-demand: Amiko provides a "Generate Doc Summary" button available within the case view. One click sends the request to Visa, and the structured summary (with citations back to the source documents) appears as a collapsible section directly below the button once ready, typically within a minute.
- VDI access: Amiko users can trigger the VDI calculation with a single click on the dispute questionnaire. The returned score helps determine whether a chargeback is likely legitimate, giving the handler a network-level signal alongside the rest of the case file before deciding how to proceed.
- Governance: Both VDI and VDDA carry a per-request fee, charged once when Visa generates the summary or score. Visa states that it stores the result on its side, so it can be retrieved as often as needed afterwards at no further cost.
By bringing VDDA and VDI data directly into the Amiko workflow, we remove the context switching and increase efficiency by unifying tools. Your dispute agents are able to see the Visa AI output in context next to everything else they need, and move from data collection to decision-making in seconds.
Frequently asked questions
What is a Visa Dispute Document Analyser (VDDA)?
VDDA is a Visa service that uses AI and Optical Character Recognition to scan documents attached to a dispute case and produce a structured summary of the critical data points. It works across all dispute categories (fraud, authorisation, processing errors, and consumer disputes) and is particularly valuable in cases with heavy evidence packs, such as first-party misuse ("friendly fraud"), where compelling evidence is buried in long authentication logs or shipping records.
How does VDDA work in practice?
When a case has attachments, agents click "Summarise with Doc Analyser" and within about 60 seconds, VDDA returns a structured report broken into sections such as Purchase Transaction Detail, Order Detail, and Service Detail. Every extracted field includes a citation to the source document, page, and attachment ID so agents can verify the output instantly.
What is the Visa Dispute Intelligence (VDI)?
VDI scores the dispute using Visa's global transaction data, returning a stage-specific win-likelihood score and key contributing factors to guide whether to submit, accept, decline, respond, or investigate further. Both are request-based, triggered by the issuer rather than running on every case.
What does VDDA cover, and where does human judgement come in?
VDDA is scoped to extract the facts from the documents. Mapping those facts to reason-code logic and assessing whether the evidence remedies the dispute under scheme rules stays with the case handler. Even in dense cases, VDDA surfaces the technical fields in seconds. Interpreting them against the cardholder's specific claim is the handler's call.
What do VDDA and VDI cost, and how do issuers enable them?
Both tools are opt-in and request-based, with a per-request fee charged only when an issuer triggers a Doc Summary or a VDI calculation (see the Visa Fee Schedule for current rates). Issuers can switch either tool on at any time via the Visa Online configuration portal or through their processor. There is no fee for simply having a tool enabled.
How are VDDA and VDI integrated into Amiko?
Both tools are available in the case view, so dispute agents don't switch systems. One click triggers a VDI score to help validate whether a dispute is likely legitimate. A "Generate Doc Summary" button returns a structured VDDA summary with citations within a minute. Both carry a per-request Visa fee, charged once at generation. Visa claims to store the result for later retrieval.