Polygrade Raises $6.3M for AI Warranty Claims

The warranty is sold before anything breaks. Its credibility is decided later, when a homeowner, carrier, contractor, supplier, and payment team all need the same claim to move through systems that were rarely designed to work together.

Polygrade has emerged from stealth with $6.3M in Seed funding to put AI agents inside that handoff. Construct Capital led the financing, while American Family Ventures, Nine Four Ventures, Hustle Fund, Room, and Pillar participated across the capital raised.

The transaction is also more precise than the headline suggests. BetaKit reports that Polygrade assembled the $6.3M through a $3.3M SAFE in early 2025 and a $3M SAFE from Construct in July 2026. The October 6 announcement brings those previously undisclosed financings into one Seed total and gives Polygrade capital to expand forward-deployed engineering, go-to-market, and partner enablement.

What Polygrade Is Building

Most claims platforms are good at maintaining a record. They can open a ticket, store policy information, route a task, and mark a claim complete. The actual work still crosses phone calls, spreadsheets, parts catalogs, contractor systems, payment queues, and judgment calls that live outside the record.

Polygrade is building an execution layer above those systems. Its task-specific agents support first notice of loss, homeowner intake, triage, authorization recommendations, parts sourcing, invoice processing, and contractor payment, while the carrier's current platform remains the system of record. The company says it connects through APIs, event streams, file exchange, or custom integrations rather than requiring a multi-year replacement project.

The boundary around automation matters in claims. Polygrade says its agents operate inside each customer's rules, make their recommendations traceable, and escalate decisions that fall outside defined thresholds for human review. That design lets a carrier automate routine work without pretending that contract interpretation, unusual damage, incomplete evidence, or an unhappy customer can be reduced to one universal model decision.

A Team Built Inside Home Services

Polygrade was founded in 2025 by David Steckel, Founder and CEO; Eui Chung, co-founder and CTO; and Kaustubh Vongole, co-founder and Head of Product. The company was initially called ProPay, reflecting an early focus on faster contractor payment, before the team expanded its scope across the full claim lifecycle and adopted the Polygrade name.

The founders' operating history is central to the investment case. Steckel founded Setter, which Thumbtack acquired, then worked on home-services businesses at Thumbtack and served as Chief Product Officer at Sears Home Services. Chung spent nearly 30 years across Sears and Transformco, rising from software developer to CTO and later leading an AI transformation inside Sears Home Services. Vongole led fulfillment products and AI strategy across a network of more than 2,500 technicians.

That experience puts the company closer to the dispatch queue than the demo stage. The founders have managed the work that happens after a claim enters the system, including routing, fulfillment, parts, customer communication, and payment. Polygrade's product thesis is that AI becomes commercially useful when it coordinates those handoffs instead of adding another interface for claims teams to manage.

Where the Operating Leverage May Live

Polygrade identifies Fidelity National Home Warranty, Cinch Home Services, Elevate Homescriptions, and United Home Warranty among its enterprise customers. Its supplier ecosystem includes Encompass, UED, Sibi, and Marcone, giving the product access to the availability and pricing decisions that determine whether the right part reaches a technician before a second truck roll becomes necessary.

The company reports 80% to 83% homeowner engagement over SMS after claim initiation, approximately 60% job-code prediction accuracy, and 7.9% of submitted claims identified as avoidable truck rolls through triage. Polygrade also says parts-order throughput can rise from six to 60 orders per agent hour, while its system identifies procurement overspend and claims where replacement may be more economical than repair. These are company-reported operating results, not independently audited benchmarks, but they show where the product is trying to earn its keep.

The economic chain is wider than one carrier's labor budget. Better triage can reduce unnecessary dispatches, earlier parts identification can shorten repair cycles, faster authorization can make warranty work more attractive to contractors, and cleaner payment can improve service-provider capacity. A platform that changes those handoffs could improve the homeowner experience without asking a carrier to replace every underlying system first.

Why Investors Are Following Warranty Infrastructure

Construct Capital's thesis starts with a category that still relies heavily on manual coordination despite the scale of the assets covered. The firm cites $2.27T in US durable-goods spending in 2025 and $30.4B in warranty claims paid by US manufacturers, figures that exclude extended and home warranties. The market is large, but the workflow remains fragmented across administrators, contractors, suppliers, manufacturers, and customers.

American Family Ventures brings insurance technology context, while Nine Four Ventures focuses on the built world. Their participation alongside Construct, Hustle Fund, Room, and Pillar reflects a bet that warranty claims sit at the intersection of vertical AI, insurance operations, field service, and physical commerce. The opportunity is not merely to classify text faster; it is to move a real repair through the people, parts, permissions, and payment required to finish it.

Polygrade plans to move beyond home warranty into extended warranty, OEM warranty, and equipment-breakdown insurance. Each category changes the underlying policy and customer relationship, but the operating pattern remains familiar: diagnose what failed, determine coverage, decide what should happen, obtain the right part or replacement, and pay the parties that performed the work.

What the $6.3M Must Prove

Forward-deployed engineering is a practical response to a market full of old systems and custom processes. It is also an expensive delivery model if every deployment becomes a bespoke integration project. Polygrade now has to show that connector patterns, task-specific agents, and customer rules can be reused enough to produce software economics while preserving the domain judgment that won the first accounts.

The company must also keep its evidence boundary visible. Claims automation can create faster decisions, but accuracy, traceability, fairness, and human escalation become more important as the software moves from recommending work to executing it. Polygrade's stated approach keeps people responsible for exceptions; the next phase will show whether that boundary remains legible as more modules and warranty categories come online.

The $6.3M gives Polygrade room to hire the engineers who connect the product to messy environments and the commercial team that turns founder-led relationships into a repeatable market. Warranty software has spent decades documenting promises after the sale. Polygrade is trying to build the layer that helps the industry keep them when something finally breaks.