Kevin Geoffrey Kemp Day, licensed insurance broker

30+ years in insurance. 25+ years in technology.

About

Kevin Day

Licensed in British Columbia. Insurance since 1994. Software since 1998. Ocean Falls Insurance Services Limited is the brokerage.

Individual licence

Level 2 (General Agent 2)

LIC-2020-0023006-R01, active

BC general insurance. In insurance since 1994.

Agency

Ocean Falls Insurance Services Limited

Agency licence LIC-2022-0037261-R01

Independent brokerage in British Columbia.

Software since 1998: .NET platforms, REST APIs, MCP servers, AI voice and chat, and CaptiveStack, on kevinday.ca. I speak your engineers' language, so the submission tells underwriters the real story of your stack and controls.

Cyber, tech E&O and AI liability lead the practice. Specialty markets and partners come in when the risk needs them.

Technology companies are the focus. Property, liability, construction, marine and personal insurance are part of the same practice.

Software

What I've built

Software for captive feasibility, and for the renewal file on a broker's desk.

CanadianCaptive

A captive feasibility pre-screen, with a domicile reference behind it.

  • Industry, premium, objective, and five-year loss ratio come back as a viability rating and a structure: single-parent, cell, or group.
  • Legislation, minimum capital, and tax notes for each domicile are generated from one dataset.

Stack · HTML, Cloudflare

TwinCover

A local assistant for a commercial renewal.

  • A local model reads the declarations and fills named insured, limits, dates, and premium. The broker confirms the fields next to the policy text.
  • It sorts the PDF: policy, binder, certificate, endorsement, or quote.
  • Coverage questions are answered from that client's documents, with the section cited. The client letter can be polished before it exports to Word or PDF.

Stack · Python, FastAPI, React, Ollama, ChromaDB

CaptiveStack

A workspace for group-captive feasibility.

  • A member survey builds the file: current policy, exposures, five-year claims, loss runs, retention, and a capital pledge.
  • Premium, loss history, and group size become a fit score, plus an illustrative five-year surplus range.
  • Association staff, members, and reviewers each open their own part of the file.

Stack · SvelteKit, Cloudflare Pages, D1

I build these systems, so I know how an AI product is assembled, where a model stops being reliable, and what to ask an underwriter.

github.com/kevingkday