Stop Manually Denying 80% of LIHTC Applicants

Property management software like RealPage and AppFolio are excellent systems of record, but their built-in application modules are too generic for LIHTC compliance. They don't have the logic to anticipate income from a pay stub with variable hours, parse a one-time bonus, or handle student status rules. This forces leasing agents to manually read every document, calculate income in a separate spreadsheet, and then tag applicants in the system. This manual process is the primary bottleneck.

Many teams resort to shared Excel or Google Sheets files to track applicants. This approach is fraught with risk. A single typo in a formula can miscalculate an entire column of applicant incomes, leading to incorrect denials and potential fair housing violations. These spreadsheets are not auditable, are difficult to maintain, and create a massive compliance liability when a single cell error can disqualify a family in need.

Trying to solve this with generic OCR tools also fails. An off-the-shelf document reader can extract text from a pay stub, but it doesn't understand context. It cannot distinguish Year-to-Date earnings from the current pay period or correctly identify a shift differential. The output is a block of unstructured text that still requires a human to interpret, defeating the purpose of automation.

Syntora would begin an engagement by integrating with your existing RealPage or AppFolio APIs to ingest new applications. We propose using the Claude API for its advanced document parsing capabilities to accurately extract structured data from diverse documents such as PDF pay stubs, offer letters, and bank statements. A critical initial phase involves deep collaboration with your compliance team to precisely codify your organization's rules for anticipating the next 12 months of income, ensuring accuracy and regulatory adherence. This custom logic would be developed in Python, leveraging libraries like httpx for resilient API communication.

The core of the proposed system would be a FastAPI service designed to orchestrate the entire workflow. Upon application submission, a webhook would trigger this service, which then manages sending documents for parsing, executing the custom income calculation engine, and applying all relevant LIHTC rules. For properties with layered programs like HOME, the system would automatically flag applications requiring full asset verification for human review. We would implement Supabase with a PostgreSQL database to maintain an immutable audit trail of every calculation, essential for compliance reporting and transparency.

Following the anticipated income calculation, the system would determine the correct AMI bucket (30%, 40%, 50%, 60%, 70%, or 80%) based on the codified rules. It would then write this data back to a custom field on the applicant's record via the RealPage or AppFolio API, and, if applicable, add the applicant to the appropriate dynamically sorted waitlist. The final component would involve configuring automated notifications, such as an email acknowledging the application and providing a projected qualification status, to manage applicant expectations and reduce manual communication burdens. The architecture would be designed for scalability and cost-efficiency, utilizing services like AWS Lambda for flexible deployment. Structured logging with structlog and robust monitoring in Datadog would be implemented to ensure high availability and operational visibility.