Fontana would sit around JBDH’s existing records, controls and exception processes, converting messy operational communications into structured, validated, approval-ready workflow records with a replayable evidence trail.
JBDH’s operating model
Internally built core
SQL Server-backed systems, browser-based internal interfaces, internal development capability and reconciliation views.
One book of record
A golden-source concept is used across reconciliations, finance, payments and related internal workflows.
Exception-led operations
Real-time matching and exception handling already exist for unmatched trades and differences.
Execution-side model
JBDH sources liquidity for clients and does not take proprietary positions. The operating model is largely STP after trade entry.
Connectivity
FIX supports a large amount of trade and counterparty flow. MarkitWire is noted for swaps, subject to transcript confirmation.
Human edge remains
Instructions and evidence still arrive through email, confirmations, Teams and Bloomberg-style channels, where interpretation and context are required.
Three hypotheses to test, not three promises
Trade amendment handling
Convert messages such as the A817 example into structured amendment evidence, validation checks and an approval-ready exception record.
Bank-confirmation comparison
Extract confirmation PDFs and perform an additive check against JBDH’s golden source or booking system. This came directly from Ben as an area of AI interest.
Recap extraction and validation
Parse richer trade recaps into structured economics, then compare currency, rate, strike, expiry, dates, notional, premium and hedge details with internal records.
Recommended next step
Choose one workflow for a 60–90 minute mapping session with Ben and the analyst or technologist who performs it day to day.
A817 / Dorman price correction
The email instruction was: “For strike 95.375, we know price 0.02. 0.2 punched.” Dorman’s response was “Fixed”.
This is a small message with a large context requirement. The operational sequence is:
- Identify the affected trade and strike.
- Compare booked price 0.2 with corrected price 0.02.
- Validate against internal records and clearer/broker evidence.
- Determine whether anything remains outstanding.
- Preserve evidence and present the proposed resolution for approval.
Why it matters
This is not a dashboard replacement. It is workflow control around an instruction that requires lookup, interpretation, validation, approval and evidence.
Human role: approve, edit or reject the proposed correction. Fontana prepares the context and evidence; it does not silently alter the book of record.
From messy trigger to controlled outcome
The JBDH-specific knowledge layer is useful as approved operational context: systems, counterparties, product conventions, owners, SOPs, approval rules and prior resolutions. It should support the workflow, not be the opening pitch.
Facts still needed before a proof of value
- Which workflow creates the most avoidable analyst work?
- Which systems and authoritative fields are checked for an amendment or confirmation?
- Who approves the action, and what evidence must they see?
- How often do amendments, breaks or recap mismatches occur?
- Where are prior exceptions and resolutions stored?
- Which clean matches could eventually be auto-filed?
- What is the safe read-only access and security-review path?
- Who owns the budget and business case?
A narrow, credible ask
Could we pick one workflow, map the trigger, extraction, source checks, exception path, approval point and audit trail, then show exactly where Fontana adds control without disrupting what JBDH has already built?
Internal working brief. The A817 and HSBC examples are retained as briefing evidence and should be source-verified before being used in a live demonstration or shared externally.