AI underwriting agent for automated loan approvals and fraud detection
Two Dots builds Eve, a conversational AI agent for loan underwriting that handles fraud detection, income verification, and document collection — tasks that traditionally require manual review. The stack (TypeScript/React frontend, Python/PyTorch backend, PostgreSQL + BigQuery for underwriting data) reflects a product-focused team moving fast on LLM-powered workflows. Active projects span document automation, financial profiling, and risk modeling, while pain points cluster around fraud detection at scale and surfacing value in atypical applications — the exact edge cases Eve is designed to resolve.
Two Dots operates an AI underwriting platform targeting loan originators and financial institutions. The product, Eve, is a conversational agent that guides applicants through underwriting by collecting documents, verifying income, and flagging fraud — delivering real-time approval decisions and reducing manual handoff. The company is based in San Francisco and currently operates at 51–200 employees with recent hiring focused on sales and data roles, though recruitment velocity is decelerating. The tech foundation spans TypeScript/React for UX, Python/PyTorch for inference, and PostgreSQL + BigQuery for data storage and analytics.
Two Dots builds on TypeScript, React, and Node.js for frontend; Python and PyTorch for ML; PostgreSQL and BigQuery for data; and GCP for infrastructure. They also integrate LinkedIn APIs and use Tableau, Looker, and Metabase for analytics.
Active projects include conversational document collection, financial profile systems, AI-driven underwriting automation, customer journey optimization, A/B testing, risk model improvement, and underwriting data analysis.
Other companies in the same industry, closest in size