Discovery & Sourcing

• Partnered with the client’s product and underwriting teams to interview potential lenders and define the right
wedge into the private credit market.
• Cataloged diligence questions across finance/accounting, business/management, legal/compliance, underwriting/servicing, and technology/infosec — linking each to its corresponding source documents (e.g., AR/AP agings, bank statements, debt schedules, org charts, insurance certificates, UCC filings, SOC/ISO audits).
• Designed document ingestion and tagging processes so the AI could understand document scope,
relationships, and context for each deal.
Architecture Build Out

•Developed a secure, GCP-hosted AI pipeline for document retrieval, OCR, classification, and structured extraction, segmented by diligence category.
•Built REST APIs returning structured answers with confidence scores and source citations down to the page level, flagging low-confidence fields for human review.
•Delivered an embeddable conversational co-pilot API, allowing underwriters to query the AI directly
(e.g., “Show AR over 60 days for April” or “Where did we get guarantor net worth?”) and cross-reference across document categories.
•Implemented conversation logging per opportunity to maintain audit trails, support compliance, and enable continuous model improvement.
Pilot & Roll out

•Provided white-glove support through the first 10 loan memos, refining parsing, field mappings, and exception workflows.
•Supplied API documentation, an onboarding run-book, partnered with the Private Credit’s front end development team to implement the APIs, and a recommended Salesforce integration architecture, ensuring Salesforce remained the single pane of glass.
•Maintained a clear division of responsibilities — client owned UI, authentication, and approvals, while XTAM focused on AI reliability and performance.
•Trained the client’s development team to manage and extend the AI infrastructure long-term, building internal capability and ownership.