Speed

Fintech

Card Deposit Management System (Shield Pay)

A unified fraud and chargeback control center that consolidates payment provider, KYC, and device-intelligence data into a single operator console.

Status

Live — in daily production use by Speed's risk team

Category

Internal fintech operations tool / risk management SaaS.

Timeline

Shipped incrementally over a few weeks (traditional build estimated at 6–9 months)

User

Speed’s risk analysts, chargeback operators, and compliance leads.

Status

Live — in daily production use by Speed's risk team

Timeline

Shipped incrementally over a few weeks (traditional build estimated at 6–9 months)

Category

Internal fintech operations tool / risk management SaaS.

User

Speed’s risk analysts, chargeback operators, and compliance leads.

THE CHALLENGE

What existed before

Before DMS, risk operators had to stitch together data from 5+ external dashboards — Checkout.com, Primer, Cybrid, Sumsub, Fingerprint, and internal Speed APIs — to investigate a single suspicious customer or dispute. Every action (restricting a card, preparing a chargeback rebuttal, verifying KYC vs. IP location) was manual, slow, and error-prone.

Slow investigations

Investigations took 30–60 minutes per case, delaying decisions on active fraud.

Manual chargeback letters

Chargeback response letters were assembled by hand in Google Docs from screenshots.

Black-box rule evaluations

Rule evaluations were a black box — operators could not see why a customer was or wasn’t blocked.

The solution

A single console for fraud and chargeback operations

DMS is a single operator console that ingests webhooks from all providers in real time, normalizes identity across accounts (via a composite customer_id / email / card BIN+last4 key), and gives analysts one-click actions backed by a full audit trail.

Key Features

Unified customer profile

transactions, KYC photos, device fingerprints, dispute history, and rule evaluations in one sheet.

Automated fraud detection

burst detection, address/name mismatch, stolen-card patterns, geo mismatch, CVV analytics.

One-click account restriction

One-click account restriction across Checkout, Primer, Cybrid (Card & ACH) with full webhook payload logging.

AI-assisted chargeback rebuttals

AI-assisted chargeback rebuttal letters and a 10-section auto-pilot PDF evidence pipeline.

Safe list / block list

Safe list / block list with dynamic AI-suggested justifications.

Integrations & tech stack

The platform is built on a modern React + TypeScript frontend, a Lovable Cloud backend with Postgres and RLS, and edge functions for secure, advisory-locked operations.

Frontend

React 18, Vite, Tailwind, TypeScript

Backend

Lovable Cloud (Postgres, RLS, Edge Functions) — server-side RPCs (decline_reason_aggregates, decline_reason_customers) for aggregation; domain-restricted auth (@tryspeed.com only) with MFA/TOTP enrollment

Integrations

Checkout.com, Primer, Cybrid, Sumsub, Fingerprint, TronGrid, internal Speed APIs

What's clever

Composite identity resolution

Accounts are unified across providers using a composite customer_id → email → card BIN+last4 key, so the same person is recognized even when each provider only exposes a partial identifier.

Lifecycle-aware webhook merging

Dispute webhooks are merged using a STATUS_RANK system instead of last-write-wins, so out-of-order webhook delivery can never downgrade a case's real lifecycle state.

Server-side aggregation, not client-side

Custom RPCs like decline_reason_aggregates push filtering and aggregation into Postgres, bypassing the 1,000-row limits and timeouts that broke client-side queries at production data volume.

Key Metrics & Impact

DMS eliminated the risk team's dependency on 5+ external dashboards, giving them one transparent, auditable console for every restriction, unblock, and evidence submission. Highlights:

Metric

Before

After

Chargeback response preparation

~60 min

~2 min per case (AI letter + auto-assembled PDF)

Account restriction time

~15 min

<10 seconds (one-click, multi-provider)

Decline analysis load time

From repeated statement timeouts

~5 seconds via server-side RPC aggregation.

Dispute list load

Serial 50-row fatches

200-row parallel fetches

Data coverage

5+ external providers unified

1,300+ safe-listed customers, hundreds of disputes tracked with full lifecycle status

Delivery vs. traditional build

Est. 6–9 months (dedicated team)

Shipped incrementally over a few weeks

Key Metrics & Impact

DMS eliminated the risk team's dependency on 5+ external dashboards, giving them one transparent, auditable console for every restriction, unblock, and evidence submission. Highlights:

Chargeback response preparation

Before

~60 min

After

~2 min per case (AI letter + auto-assembled PDF)

Account restriction time

Before

~15 min

After

<10 seconds (one-click, multi-provider)

Decline analysis load time

Before

From repeated statement timeouts

After

~5 seconds via server-side RPC aggregation.

Dispute list load

Before

Serial 50-row fatches

After

200-row parallel fetches

Data coverage

Before

5+ external providers unified

After

1,300+ safe-listed customers, hundreds of disputes tracked with full lifecycle status

Delivery vs. traditional build

Before

Est. 6–9 months (dedicated team)

After

Shipped incrementally over a few weeks

Want to build something similar?

Let’s talk about how a unified console and wallet-native engagement layer can transform your customer operations.

Want to build something similar?

Let’s talk about how a unified console and wallet-native engagement layer can transform your customer operations.

Want to build something similar?

Let’s talk about how a unified console and wallet-native engagement layer can transform your customer operations.