Help & Support

For most enterprise clients, the standard API-based integration takes 2–5 business days from contract signature to first live agent decision. The OPAL platform exposes a REST API and WebSocket stream that map directly onto common banking core and trading infrastructure patterns.

We provide a dedicated integration engineer during onboarding who will configure the agent pipeline to match your event schema, authentication model, and delivery endpoints. On-premise deployments in isolated environments may require 2–4 additional weeks depending on your infrastructure review and security sign-off process.

By default, OPAL operates as a stateless processing layer. Transaction event payloads are processed in memory and not persisted to OPAL infrastructure beyond the configurable short-term audit buffer (default: 72 hours, configurable down to zero for on-premise deployments).

For clients requiring zero data egress, OPAL offers a fully containerised on-premise deployment where all inference, logging, and storage occurs within the client's own security perimeter. Under this model, no transaction data leaves your infrastructure at any point.

The OPAL Compliance Agent ships with pre-configured rulesets for the following frameworks:

• CBR (Central Bank of Russia) — Forms 550, 634, 0409251
• FATF Recommendations — 40 + 9 special recommendations
• MiFID II — transaction reporting and best execution logging
• Federal Law No. 115-FZ (AML/CFT) — Russian AML requirements
• SWIFT MT/MX message compliance checks

Custom regulatory rulesets for other jurisdictions (e.g., EU DORA, SEC Rule 17a-4) can be added as a professional services engagement. Contact [email protected] to discuss requirements.

Every agent decision includes a structured explanation record with a confidence score, the specific signals that drove the decision, and a unique decision ID for traceability. Clients configure approval thresholds per agent — decisions below the threshold are automatically escalated to a human analyst rather than acted upon.

For executed decisions that are subsequently found to be incorrect, the platform provides a correction workflow: the decision is logged as overridden, the correction is applied, and the event is fed back into the model's online learning buffer (if enabled) to improve future accuracy. All corrections are retained in the immutable audit log.

Yes. Qualified prospective clients receive access to a sandboxed OPAL environment as part of the demo and evaluation process. The sandbox is pre-loaded with synthetic financial event data representative of real production volumes, allowing you to evaluate agent latency, decision quality, and integration effort without using live data.

Sandbox access is available for a 14-day evaluation period following a scoping call with our team. To request access, contact [email protected] or use the demo request form on the main site.

OPAL is priced on an enterprise subscription basis, scoped to the number of agents deployed, monthly event volume, and deployment model (cloud-hosted vs. on-premise). There are no per-decision fees — you purchase capacity, not consumption.

Indicative pricing is provided during the demo process after we understand your event volumes and agent requirements. All pricing is in RUB or USD depending on client preference and jurisdiction. Contact [email protected] to discuss your requirements.