Problem and Solution

The Problem

Mortgage loan origination remains fragmented and manual across borrower intake, document collection/verification, AUS submissions (DU/LPA), condition clearing, and handoffs among loan officers, processors, and underwriters. Legacy LOS platforms are indispensable for compliance and investor delivery but are often rigid, create duplicated effort, and require many “touches per loan.” The result is high cost per loan (often ~$10K‑$11K per loan for IMBs per MBA data), long cycle times, and frequent defects/rework.

Coevolved’s Solution: Porchlight

Coevolved is building Porchlight, an agentic automation layer and POS that sits on top of existing LOS systems. It guides borrowers through applications, answers questions, and automates document collection and verification. It connects to AUS systems (DU/LPA) and external data providers to verify borrower information automatically. For ops teams, Porchlight orchestrates repetitive back‑office work like condition clearing, status updates, and handoffs—keeping humans in control and maintaining auditability.

Value Proposition and Differentiation

• Faster cycle times and fewer touches per loan through automated doc collection, verification, and condition management.
• Reduced defects and rework via consistent workflows and clear audit trails.
• Human‑in‑the‑loop control with observability to meet compliance needs while leveraging AI agents.
• Integrates with lenders’ existing LOS and data services rather than forcing rip‑and‑replace.
• Potential managed inference and multi‑agent orchestration capability informed by the team’s open‑source agent framework work.

Impact

If Porchlight reduces opex even 5‑10% at scale, lenders could save millions annually and improve throughput. Better borrower experience should improve conversion rates and NPS while reducing fallout. The platform aims to make AI‑driven automations production‑ready with traceability suitable for regulated environments.

Market and Competitors

Market Context and Trends

• Mortgage tech budgets are shifting toward digital lending platforms and AI‑enabled automation to shrink cycle times, reduce cost per loan, and improve borrower experience. Research indicates rapid growth in digital lending platforms globally (CAGR ~27.7% to 2030) with a sizable U.S. market.
• Lenders rely on a core LOS (e.g., ICE Encompass, MeridianLink Mortgage, LendingPad) plus borrower‑facing POS (e.g., Blend, Maxwell, BeSmartee, Floify) and many verification/eClose integrations.
• AI copilots/agentic workflows are emerging in POS and back‑office operations, but auditability and integration depth are key adoption barriers that vendors must address.

Competitive Landscape

Core LOS vendors (incumbent backbone, indirect competition/partners)

• ICE Mortgage Technology (Encompass LOS)
• MeridianLink Mortgage (LOS)
• LendingPad (LOS)
• Calyx (Point/Path) (LOS)

Mortgage POS and digital origination suites (direct competition for borrower journey and some workflow automation)

• Blend (enterprise POS + verifications + closing)
• Maxwell (POS + eClose)
• BeSmartee (configurable POS)
• Floify (borrower portal/doc collection; integrates with LOS)
• SimpleNexus (mobile‑first digital mortgage; nCino)
• Blue Sage (end‑to‑end digital lending with POS/LOS components)
• LenderLogix, Lenderful, Arive, Loanzify, Pre‑Approve Me, PreApp 1003, Take3Tech

Workflow/task automation and modular LOS alternatives (overlap in orchestration, doc handling, and conditions)

• nCino Mortgage Suite (with SimpleNexus) for banks/CUs
• Tavant (Touchless Lending; LO.ai)
• TaskSuite and other modular workflow‑centric LOS/automation tools

Positioning for Coevolved

• Integrate‑with‑the‑LOS strategy: By sitting atop existing LOS and focusing on agentic orchestration, Coevolved can complement entrenched systems while competing with POS/automation suites on speed of deployment, automation depth, and observability.
• Differentiators: agentic, composable architectures; human‑in‑the‑loop auditability; focus on repetitive ops work (condition clearing, status orchestration) that POS tools sometimes under‑serve; open‑source agent framework pedigree.
• Risks: Procurement sensitivity to AI in regulated workflows, need for robust compliance and vendor management, and competition from incumbents expanding AI features. Success depends on demonstrable ROI, integration breadth (DU/LPA, verifications, eClose), and referenceable lender wins.

Founders and Leadership Background

• Founders: Athan Zhang (Founder/CEO), Alex Li (Co-founder)
• Company: Coevolved (YC W26), founded 2025
• Website: https://www.coevolved.ai

Athan Zhang (Founder & CEO)

Athan Zhang is the founder and CEO of Coevolved, a Y Combinator‑backed startup building AI‑powered automation for mortgage loan origination. His personal site notes he studied Computer Science at Princeton University and left to join YC, underscoring a strong technical background and entrepreneurial drive. His experience includes MLOps and quantitative development internships (Leidos; Five Rings) and prior leadership of Princeton Student Ventures. Before university, he conducted deep learning research at George Mason University (continual learning), ran a math/physics prep course, and medaled in Science Olympiad competitions. He also co‑started Vytal, an early medtech team that reportedly raised $1.2M. Outside work, he is active in endurance sports and mentoring.

Education: Princeton University (Computer Science; departed to found Coevolved via YC W26). Notable achievements: YC‑backed founder; prior medtech venture involvement; internships in quantitative finance and MLOps; leadership in student entrepreneurship.

Alex Li (Co‑founder)

Alex Li is listed as a cofounder of Coevolved (YC W26). His LinkedIn public profile confirms his role and indicates education at the University of Washington (2019‑2023 listed as schooling years in the visible portion; additional details are gated). The YC company listing also names Alex Li as a founder. While detailed professional history is not fully accessible without login, available sources corroborate his cofounder status and technical orientation through Coevolved’s open‑source GitHub presence and the company’s product focus on agentic automation.

Education: University of Washington (per LinkedIn public profile snippet; specifics and degree not fully visible). Notable achievements: YC‑backed cofounder; co‑maintainer of Coevolved’s agent framework repo.