Book a demo

Platform

One control plane for every dispute

Connect payment, order and fulfilment data, build the evidence for each case, recommend the action and automate the approved workflow.

Book a demoSee pricing

Architecture

Inputs, control plane, outputs

The platform wraps around existing infrastructure. No rip and replace.

Inputs
  • PSPs, acquirers and gateways
  • CRM and customer support
  • Orders, ERP and fulfilment
  • Email, forms, chat and calls
  • Booking, ticketing, GDS and PMS
SIMPLIChargeback control plane
  • Dispute and transaction data model
  • Secure integration fabric
  • Evidence graph and document store
  • Scheme rule and deadline engine
  • AI orchestration and permissions
  • Human review queues
  • Reporting and audit ledger
Outputs
  • Accept or refund recommendation
  • Prevention or deflection action
  • Evidence package and representment
  • Customer communication
  • Processor submission
  • Finance and compliance reporting

Workflow

From notification to outcome

Five stages, each with a clear owner and a clear record.

  1. Ingest and classifyDisputes are deduplicated and mapped to the right reason code and deadline.
  2. Retrieve evidenceAuthorised records are pulled into one evidence graph.
  3. Score and decideCompleteness and expected net recovery decide fight, accept or refund.
  4. Draft and reviewA source cited response is drafted and routed by threshold.
  5. Submit and learnCases are filed on time and outcomes improve the next case.

AI agents

A team of specialist agents, with people in control

Each agent does one job well, and none can act outside the permissions you set.

  1. IngestDisputes arrive from every processor and are deduplicated.
  2. UnderstandReason code, deadline and context are mapped.
  3. Build evidenceAuthorised records are retrieved and checked for gaps.
  4. Decide and draftExpected net value decides the action, and a cited response is drafted.
  5. Review and submitApproved cases are filed before the deadline, with a full audit trail.
Learning loop. Outcomes feed back into retrieval, rules and evidence selection.
Ingest agentReceives, normalises and deduplicates disputes.
Classification agentMaps reason codes, narratives and transaction context.
Evidence agentRetrieves authorised payment, order, fulfilment and CRM records.
Contradiction agentFinds missing, conflicting or unsupported evidence.
Decision agentPredicts recovery and calculates expected net value.
Drafting agentPrepares source cited, reason code specific responses.
Review agentApplies confidence, value, novelty and policy thresholds.
Submission agentSubmits approved cases through APIs or assisted workflows.
Audit agentRecords inputs, sources, model versions, approvals and outcomes.
Learning loopUses outcomes to improve retrieval, rules and evidence selection.
Human control. People stay involved in high value, ambiguous, novel, service failure, arbitration and regulatory sensitive cases.

Platform features

Everything you need to connect, decide and control

Twelve capabilities, grouped by the job they do.

Connect

  • Multi PSP ingestionDisputes from every processor in one canonical model.
  • Secure APIs and file connectorsWraps around existing systems.
  • Gateway and ecommerce connectorsPayment gateways, marketplaces and Shopify.
  • Model agnostic AI connectorsUse the language model you approve.

Decide

  • Reason code classificationVisa and Mastercard codes, with explanations.
  • Evidence scoringCompleteness and quality checked before filing.
  • Scheme rules and deadlinesA versioned library and deadline engine.
  • Net recovery prioritisationEffort goes where value is.

Control

  • Source cited draftingEvery claim points to its record.
  • Approval gatesReview queues set by your thresholds.
  • Controlled submissionRepresentment through APIs or assisted flows.
  • Reporting and audit trailCase management, analytics and a full ledger.

Security and compliance

Autonomous preparation, controlled execution

Sensitive payment data is handled with care, and every automated step can be explained to an auditor.

  • Obfuscated PAN. Card numbers are masked, so agents work without exposing full account numbers.
  • Permissioned agents. Each agent can only read and do what you have explicitly allowed.
  • Human approval gates. High value, ambiguous, novel and regulatory sensitive cases go to a person.
  • Full audit trail. Inputs, sources, model versions, approvals and outcomes are recorded for every case.
  • Source cited drafting. No unsupported claims. If the evidence is missing, the response says so.
  • Model agnostic. Bring the language model your security team has approved.

Immutable audit ledger

Card5412 75** **** 4417
Evidence retrievedOrder, fulfilment, CRM
Model versionRecorded
ConfidenceAbove threshold
Approved byNamed reviewer
SubmittedBefore deadline

Next step

See SIMPLIChargeback working on your own disputes

Share a sample of historical disputes and we will show you the evidence, the recommended action and the audit trail.

Book a demoView pricing