Saphyre

Aigent reads account details and documents arriving by email or chat, validates and classifies them and chases missing information, and the platform normalizes each firm's data. The permissioned shared repository, status tracking and workflow underneath would keep working on its own.

Maker and checker rules apply multi level verification, and a live dashboard shows which party owes the next step. Aigent can push data into a firm's systems by API with no person touching it, and the validations before parsed data reaches a counterparty are not described.

Every action is logged in real time with a timestamp and the person or agent responsible, agent tracing is available on request, and 108 patents describe the method. Accuracy for parsing and model documentation stay unpublished.

BlackRock's Liliane Ancona says hours of work became minutes, and BNY Mellon's global head of custody, Caroline Butler, says days or weeks became near real time. Executives at JPMorgan, Northern Trust and Citi describe production use, and the figures are quotes rather than measured results.

Information reaches only the counterparties a firm chooses, agents retain no memory, and AI features run only when a firm opts in. Whether data structured for one firm informs processing for another is not stated.

The privacy policy covers website visitors and cookies, while the platform holds KYC files, tax documents and settlement instructions for funds whose managers compete. Data processing terms, retention and a subprocessor list for the platform are not published.

SOC 2 Type II compliance and third party penetration testing are listed beside single sign on, multifactor authentication and IP whitelisting. Audit period and report access are not given.

Readiness for the T+1 settlement cycle, exportable messaging for electronic communications compliance and questionnaires updated as rules change form the stated position. No supervisory instrument beyond the settlement cycle appears.

A misparsed account detail or wrong standing settlement instruction produces a failed trade or a payment to the wrong place. Parsing accuracy across document types and market specific forms, and how an ingestion error is caught before it spreads, are unpublished.

Action ownership lets counterparties trace a delayed or wrong onboarding to the party that owed the step. Remediation for a misread settlement instruction, or any accuracy guarantee, is not published, and an error made once spreads to every permissioned counterparty.

Proprietary intelligence keeps the analytical layer owned rather than licensed, and integration partners are named individually. Neither the model reading Aigent's emails nor the hosting provider is named.

FXall and 360T give direct access to account workflows and statuses, DocuSign handles signatures and GLEIF supplies legal entity details, with Teams and Symphony for messaging. Entered once, data is available to every permissioned counterparty across a network of more than 40 institutions.

Storage and processing may happen wherever partners or service providers have facilities, per the privacy policy. Hosting provider, region and separation between competing counterparties are left unstated.

Value accrues to every counterparty on a shared platform, yet whether the buy side, sell side or custodian carries the cost is never said. Modules such as legal agreement management and readiness monitoring are unpriced too.

Investment managers, brokers and custodians each have role pages, with a solution page per workflow from trading account management to securities funding. Its boundary is institutional securities operations.