How does the assistant reconcile trades between multiple brokers and the general ledger?
The assistant normalizes trade events from each source into a single timeline, matches trades to ledger events using order IDs, timestamps and notional amounts, and surfaces unmatched rows with suggested ledger mappings and variance notes. Controllers can review suggested journal entries, edit charts of accounts, and export reconciliation rows as CSV/JSON for posting.
What data sources and file formats are supported for FX trade and price feeds?
Common FX sources are supported: MT4/MT5 logs, OANDA/FXCM/IG/Dukascopy broker blotters, Interactive Brokers statements, FIX/OMS streams, market tick sources, CSV/Excel blotters, GL extracts, SQL warehouses, and S3/SFTP dumps. Start with file-based ingestion and add live API or streaming feeds as needed.
How are AI suggestions documented so auditors can reproduce decisions?
Every suggestion includes the original prompt, the exact input files or feed slices used, a plain-language reasoning summary, and a stamped decision log that records who reviewed or approved the suggestion. Audit exports preserve these fields so auditors can trace recommendations back to source events.
Can I enforce custom risk rules and have them trigger alerts in real time?
Yes. Configure position limits, cross-currency exposure thresholds, and margin-utilization rules. Rules run against the unified timeline and can trigger operator alerts with recommended mitigations and an execution checklist for traders and ops teams.
How does the system detect feed problems, slippage, or duplicate executions?
Feed-health sensors monitor latency spikes, missing ticks, out-of-sequence messages and price anomalies. Trade comparisons flag slippage and duplicate executions by matching timestamps, sequence numbers and price deltas, then open an incident with annotated timelines and suggested remediation steps.
What control do finance teams have over automated journal entry suggestions?
Finance teams can review, edit, and approve suggested journal entries in a dedicated reconciliation workflow. Suggested mappings include rationale and variance notes; controllers can attach supporting evidence, require additional approvals, and export signed entries for posting.
How long does it take to connect a typical broker blotter and start receiving reconciliations?
Time to value depends on source accessibility. File-based blotters (CSV/Excel) can be ingested and reconciled within hours once mapping rules are configured. Live API or FIX onboarding requires broker credentials and mapping checks but follows the same normalization and reconciliation workflows.
How is sensitive trade and client data protected during ingestion and storage?
Data is ingested over encrypted channels and stored according to configured retention and access-control policies. Role-based permissions restrict who can view PII and trade-level details, and exports preserve audit metadata to ensure traceability without exposing unnecessary client data.
Does the assistant provide backtest summaries that are reproducible and explainable?
Yes. Backtest reports include the exact tick files and parameter sets used, a plain-language summary of strategy behavior, drawdown windows, and a data provenance section describing any gaps or adjustments made to the input data.
How do traders and accounting teams collaborate on exception remediation and sign-off?
Exceptions open a shared incident with an annotated timeline, operator notes, suggested corrective entries, and an approval field. Traders, ops and controllers can add comments, attach evidence, and complete sign-off steps; the final signed record is included in the audit export.
