Payments & Clearing Execution
The 4 screens that manage smart payout routing, chargeback and dispute defense, settlement and clearing, and financial escrow.
What this group does
Payments & Clearing Execution is the final group in the Financial Audit & Fintech Ops add-on, covering smart payment routing, chargeback/dispute management, acquirer settlement batches, and contractual escrow. With this group, all 16 operational screens across the add-on are implemented. All 4 screens use the identical direct-per-record two-stage hybrid engine (Stage 1 embedding infrastructure, unwired to the UI; Stage 2 AI-engine-routed synthesis) described on the Overview page.
Screen preview
A quick look at the screen. Use the toggle in the corner to switch between the dark and light themes.
Settlement, clearing and escrow status
1. Payout Routing Rules
Purpose: A smart-routing ledger tracking a payment routing rule’s target gateway/rail (Stripe, Adyen, FAST/EFT, SWIFT, etc.), priority tier, fee overhead, and success rate — lets payment teams watch overall gateway approval-rate health and trigger an emergency PSP failover when a rail’s failure rate spikes.
Ledger columns / key fields
- ID format
PRR-YYYY-XXXX ruleId,targetGatewayOrRails,priorityTier,feeOverheadPct,successRatePct,maxSingleTicket,currencyroutingStatus(active_route/failover_active/disabled),countryCode,lastRunDate,aiRoutingDiagnostic,notes
Gauge
Routing Efficiency Gauge — the average success rate across records that are not disabled only — disabled routes are excluded from the average entirely, not counted as zero. Returns 0 with no records.
Key actions
- New Record, AI Analyze (runs the payout routing audit, producing a routing diagnostic plus confidence score; disabled while target gateway/rail is blank).
- Force Gateway Failover Switch / Trigger Failover (row action, inline mini-form: fallback gateway) — updates the target gateway to the new fallback gateway, sets the routing status to failover active, and appends the change (old gateway → new gateway) to the notes — a real, direct update to this record’s own fields.
- Delete — permanent, with inline confirmation.
Known limitation
Trigger Failover does not redirect real live transaction traffic. There is no payment-gateway orchestration/execution system anywhere in the app today. The action does genuinely update this record’s target-gateway and routing-status fields as the system of record for the failover decision, but no real cutover of live traffic occurs.
2. Chargeback & Dispute
Purpose: A chargeback ledger tracking a card-scheme dispute case (fraud, unrecognized charge, item not received, other) against its disputed amount, defense due date, and resolution status — tracks the portfolio-wide representment win rate and ensures evidence packs are prepared before the defense deadline.
Ledger columns / key fields
- ID format
CBD-YYYY-XXXX disputeId,cardSchemeRef,merchantOrAccount,disputedAmount,currencyreasonCode(fraud/unrecognized_charge/item_not_received/other),aiWinProbabilityPct,defenseDueDatedisputeStatus(needs_evidence/representment_submitted/won/lost),countryCode,lastRunDate,aiDisputeDiagnostic,notes
Gauge
Portfolio Dispute Win Ratio Meter — Excludes unresolved disputes from both numerator and denominator: cases still needing evidence or already submitted for representment are not counted at all. The formula is strictly won cases divided by (won + lost) cases, times 100. Returns 0 if there are no resolved (won+lost) records. This is a deliberate design choice: counting cases that haven’t been decided yet would dilute the true representment performance with cases still in flight.
Key actions
- New Record, AI Analyze (runs the chargeback/dispute audit, producing a dispute diagnostic plus win-probability score; disabled while card scheme reference or merchant/account is blank).
- Submit Representment Pack (row action, inline mini-form: evidence summary + document references) — sets the dispute status to representment submitted and logs the evidence summary to the notes.
- Delete — permanent.
Known limitation
Submit Representment Pack does not transmit any document to a real acquirer or card network. There is no acquirer/gateway integration or document-dispatch system anywhere in the app today. The action does genuinely update the dispute status to representment submitted and log the evidence summary as the system of record, but no real file transmission occurs.
3. Settlement & Clearing
Purpose: A clearing ledger tracking a settlement batch from an acquirer or card network (Stripe, Visa Direct, Mastercard Send, etc.) against gross volume, reserve deductions, and processing fees — tracks the portfolio-wide reconciliation index and enables releasing audit holds for payout.
Ledger columns / key fields
- ID format
SET-YYYY-XXXX - Batch ID, acquirer/network, gross volume, reserve deductions, processing fees, currency, clearing cutoff date
- Settlement status (cleared and paid / fee variance hold / pending clearing), country code, last-run date, AI settlement diagnostic, notes. Net Payout is derived: gross volume minus reserve deductions minus processing fees, shown in red when negative.
Gauge
Settlement Reconciliation Index — Standard status-count pattern: the percentage of records marked cleared and paid out of the total, rounded. Returns 0 with no records.
Key actions
- New Record, AI Analyze (runs the settlement/clearing audit, producing a settlement diagnostic plus confidence score; disabled while acquirer/network is blank).
- Release Settlement Hold (row action, inline Yes/No confirmation) — directly updates the settlement status to cleared and paid and appends “Settlement hold released — cleared for payout.” to the notes; disabled once already cleared and paid.
- Delete — permanent.
Known limitation, with a real-update exception
Release Settlement Hold does not trigger a real bank transfer — there is no funds-disbursement/payout-execution service anywhere in the app today. However, this action is a genuine, direct update to the record’s own settlement-status field, just as with Recalibrate Credit Ceiling and Reallocate HQLA Reserves in the Treasury & Liquidity Operations group — the status field itself is real, only the underlying bank transfer is not implemented.
4. Financial Escrow
Purpose: The final screen in this group and in the add-on. An escrow ledger tracking funds held in trust for a buyer/beneficiary against a contractual release condition or milestone — tracks the portfolio-wide trust capital reserve and enables executing escrow payouts once conditions are verified.
Ledger columns / key fields
- ID format
ESC-YYYY-XXXX - Escrow ID, buyer/beneficiary, contract reference, locked reserve amount, currency, release condition, expiry date
- Escrow status (locked in trust / milestone verified / disputed hold / released), country code, last-run date, AI escrow verification, notes
Gauge
Trust Capital Reserve Bar — Capital-weighted, explicitly not a record-count ratio: the sum of the locked reserve amount across records that are still locked (not yet released), divided by the sum of the locked reserve amount across all records, times 100. Returns 0 with no records or zero total locked amount.
Why capital-weighted?
Escrow accounts vary enormously in size, so a naive record-count ratio would mislead. Concrete example from the QA source: a portfolio with one $1.8M released escrow and one $200K escrow still locked in trust is 2 records — a naive count-based ratio would read as 50% still locked. The real, capital-weighted formula instead yields 10% ($200K / $2.0M total), correctly reflecting that the large majority of capital has already been released — weighted by the dollar amount still held in escrow, not by how many accounts are still locked, so one large still-locked account matters more than several small released ones.
Key actions
- New Record, Run Verification / AI Doğrulama (runs the financial escrow audit, producing an escrow verification plus confidence score; disabled while buyer/beneficiary or contract reference is blank).
- Execute Escrow Payout (row action, inline mini-form: authorizing signature) — directly updates the escrow status to released and logs the authorizing signature reference to the notes; disabled once already released.
- Delete — permanent.
Known limitation, with a real-update exception
Execute Escrow Payout does not move real money out of a trust account — there is no trust-account/banking-rail funds-release system anywhere in the app today (same category of gap as Release Settlement Hold, Execute Liquidity Sweep, and Auto-Generate FX Hedge Order in their respective domains). As with Release Settlement Hold, this action is a genuine, direct update to the record’s own escrow-status field — only the underlying fund transfer is not implemented.
AI Assistant — Payout Routing Rules
The Payout Routing Rules screen ships with an embedded AI Assistant — the same right-edge panel used across TotalApp. Open it from the vertical AI Assistant tab on the right of the screen; it understands what is currently on your screen and answers in Markdown, matching the language you write in (Turkish or English).
This chat assistant is a separate, complementary feature from the screen's existing per-record AI Analyze button — AI Analyze runs a fixed routing diagnostic on one rule and returns a result for that record, while this assistant is an open-ended conversation across every route in the portfolio that you can steer, ask follow-up questions in, and use to explore results AI Analyze already surfaced. It is tuned for portfolio-wide routing review:
- Success-rate degradation — explains why a gateway or rail's success rate is degrading.
- Failover priority recommendations — recommends failover priority when a route is underperforming.
- Fee vs. success trade-offs — summarizes fee overhead against success-rate trade-offs.
- Low-priority, high-failure routes — flags low-priority routes with high failure rates.
Tick one or more rows in the table to scope the conversation, then use Analyze selected or Summarize selected. Any answer can be saved via Save as Report (it lands in My Reports), and Add Knowledge lets you attach reference documents to ground the reply.
Engine-aware
The assistant respects your Settings → Agentic writer engine: Ollama / local-LLM / in-browser models run entirely on your machine, while API routes through the server — the panel never leaks your data to a server the selected engine can’t reach.
AI Assistant — Chargeback & Dispute
The Chargeback & Dispute screen ships with an embedded AI Assistant — the same right-edge panel used across TotalApp. Open it from the vertical AI Assistant tab on the right of the screen; it understands what is currently on your screen and answers in Markdown, matching the language you write in (Turkish or English).
This chat assistant is a separate, complementary feature from the screen's existing per-record AI Analyze button — AI Analyze runs a fixed dispute diagnostic on one case and returns a win-probability score for that record, while this assistant is an open-ended conversation across the whole dispute queue that you can steer, ask follow-up questions in, and use to explore results AI Analyze already surfaced. It is tuned for portfolio-wide dispute triage:
- Defense-deadline prioritization — prioritizes disputes nearest their defense-due-date across the whole queue.
- Win-probability trends — summarizes win-probability trends by reason code.
- High-value disputes needing evidence — flags high-value disputes that still need urgent evidence.
- Status distribution by merchant — explains the dispute-status distribution by merchant.
Tick one or more rows in the table to scope the conversation, then use Analyze selected or Summarize selected. Any answer can be saved via Save as Report (it lands in My Reports), and Add Knowledge lets you attach reference documents to ground the reply.
Engine-aware
The assistant respects your Settings → Agentic writer engine: Ollama / local-LLM / in-browser models run entirely on your machine, while API routes through the server — the panel never leaks your data to a server the selected engine can’t reach.
AI Assistant — Settlement & Clearing
The Settlement & Clearing screen ships with an embedded AI Assistant — the same right-edge panel used across TotalApp. Open it from the vertical AI Assistant tab on the right of the screen; it understands what is currently on your screen and answers in Markdown, matching the language you write in (Turkish or English).
This chat assistant is a separate, complementary feature from the screen's existing per-record AI Analyze button — AI Analyze runs a fixed settlement diagnostic on one batch and returns a result for that record, while this assistant is an open-ended conversation across multiple batches that you can steer, ask follow-up questions in, and use to explore results AI Analyze already surfaced. It is tuned for portfolio-wide settlement review:
- Cross-batch reconciliation gaps — explains settlement reconciliation gaps across multiple batches.
- Unusual deduction/fee patterns — flags batches with unusual reserve-deduction or fee patterns.
- Status by acquirer/network — summarizes settlement status broken down by acquirer or network.
- Age-based prioritization — prioritizes unresolved batches by how long they have been open.
Tick one or more rows in the table to scope the conversation, then use Analyze selected or Summarize selected. Any answer can be saved via Save as Report (it lands in My Reports), and Add Knowledge lets you attach reference documents to ground the reply.
Engine-aware
The assistant respects your Settings → Agentic writer engine: Ollama / local-LLM / in-browser models run entirely on your machine, while API routes through the server — the panel never leaks your data to a server the selected engine can’t reach.
AI Assistant — Financial Escrow
The Financial Escrow screen ships with an embedded AI Assistant — the same right-edge panel used across TotalApp. Open it from the vertical AI Assistant tab on the right of the screen; it understands what is currently on your screen and answers in Markdown, matching the language you write in (Turkish or English).
This chat assistant is a separate, complementary feature from the screen's existing per-record Run Verification button — Run Verification runs a fixed escrow verification on one account and returns a result for that record, while this assistant is an open-ended conversation across the whole escrow portfolio that you can steer, ask follow-up questions in, and use to explore results Run Verification already surfaced. It is tuned for portfolio-wide escrow review:
- Approaching-expiry flags — flags escrow accounts nearing expiry with their release condition still unmet.
- Locked-reserve exposure — summarizes locked-reserve exposure broken down by beneficiary.
- Follow-up prioritization — prioritizes accounts that most need release-condition follow-up.
- Status distribution — explains the escrow status distribution across the portfolio.
Tick one or more rows in the table to scope the conversation, then use Analyze selected or Summarize selected. Any answer can be saved via Save as Report (it lands in My Reports), and Add Knowledge lets you attach reference documents to ground the reply.
Engine-aware
The assistant respects your Settings → Agentic writer engine: Ollama / local-LLM / in-browser models run entirely on your machine, while API routes through the server — the panel never leaks your data to a server the selected engine can’t reach.
Gauge diversity across this group
This group intentionally uses four different gauge formulas, none of which share the other’s pattern: Routing Efficiency is a plain average (excluding disabled records); Dispute Win Ratio excludes unresolved cases from both numerator and denominator; Settlement Reconciliation Index is the add-on’s standard status-count ratio; Trust Capital Reserve is capital-weighted. This mirrors a broader lesson across the whole add-on — Transaction Audit gauges are mostly status-count, the Treasury group’s LCR gauge is intentionally unclamped above 100%, and its Concentration Meter is weighted toward the single largest position. No gauge in this add-on can be assumed to follow a “default” pattern — always check what each gauge actually measures.