case_study / qlo
An underwriting workstation, inherited demo to production beta in four months
Jan 2026 – May 2026
Qlo — AI-native insurtech (YC S26)
Commercial insurance carriers process submissions across legacy tools and spreadsheets; Qlo consolidates that work into one AI-native workstation. As the sole frontend engineer, I inherited a demo-stage frontend with no handover and took it to a production beta.
demo → beta
in ~4 months
quote → bind
end-to-end workflow
sole
frontend engineer
The hard part
Underwriting ends in documents that carry legal weight: quotes, binders, invoices. Every carrier client wants its own variation of the quote form, a single field change can reprice the policy, and what reaches the broker has to match exactly what was priced.
What I built
Extended the submission workflow from quote to bind, adding custom quote options, binders, and invoices. Designed a schema-driven quote form (React Hook Form + Zod) that handles per-client variations, and improved the premium calculation behind it. Outside the core workflow, I redesigned the login flow with 2FA and built broker management.
Decisions that mattered
Per-client quote schemas lived in the frontend; I proposed serving them from the backend as a contract, so onboarding a new client becomes configuration rather than a frontend release. I also proposed rendering quote letters as PDF instead of HTML, so every recipient sees the same fixed document and the issued version can be retained as-is for compliance — and the team shipped it that way. Separately, I brought engineering into design reviews before spec sign-off, so implementation constraints surfaced while the spec was still cheap to change; the team adopted it.
Outcome
From an inherited demo to a production beta in roughly four months, tested by a design partner whose feedback informed ongoing iterations. The product is internal and non-public, so this page shows the architecture rather than screenshots — the diagram below is the shape of the submission workflow.
