To our knowledge, Seaworthy Insurance is the first disability insurance brokerage to expose an agent-callable quote action over MCP. An AI assistant, with its user's permission, can connect to our knowledge base, pull verified carrier facts, and submit a real quote request on a professional's behalf. The request goes straight from the assistant to a licensed advisor, with no form and no phone call on the way. It has been live and publicly listed since early June.
Why can AI assistants suddenly reach outside services?
The Model Context Protocol (MCP) is the open standard AI assistants use to connect to outside tools and services. Think of it as a portal on a website that agents can connect to and act through. Without one, they are limited to reading whatever is published.
On July 28th the MCP plumbing changed, swapping a two-step session "handshake" for plain single calls, which makes it far easier for your assistant to reach out and transact with the real world. Plumbing can be, well, not the most interesting development, but in this case it earns attention. It removes the last technical excuse for consumer AI assistants to gatekeep their users.
Are people actually buying insurance through AI assistants?
For commodity lines, yes, and more so than people probably realize. Experian launched an insurance marketplace app inside ChatGPT in February 2026, comparing auto insurance directly in the chat window. OpenAI's Agentic Commerce Protocol, developed with Stripe, allows people to purchase inside the ChatGPT UI.
Auto and home insurance are the first candidates for this because they are easily legible to a machine. The rates are published and there are only a handful of coverage choices, meaning an agent can shop them because there is nothing to truly compare other than price.
I suggested in June that commodity insurance would move into the chat window, whereas individually underwritten coverage would not (yet, at least). Seven weeks later that appears to be holding.
Will an AI agent buy your disability policy end to end?
No, because of the medical file. Individual disability coverage is priced on your occupation and your income, and then on whatever your medical history turns up. Offers are frequently modified depending on the individual's unique circumstances.
Roughly 28% of the policies in our placed book were issued with an exclusion rider or a rating attached, per the 2026 audit. Those modifications get negotiated, sometimes away entirely, and that negotiation runs on medical evidence and human judgment. An agent cannot yet negotiate whether an exclusion rider is acceptable based on precedent or push for it to be reconsiderable in two years.
For example, an applicant with a documented history of anxiety or depression may get an offer capped at a five-year benefit period. If the carrier has also attached a mental/nervous exclusion, we argue that the two are doing the same job. The condition is already carved out of the contract. A shorter benefit period reaches the claims that have nothing to do with it, such as a cancer diagnosis or a spine injury twenty years out. Carriers do not always agree. We treat that first offer as a starting position.
So what is a quote action for, if not buying?
People are already researching insurance with their AI of choice. The answer is assembled from whatever that model happens to retrieve, which can vary from one to another. When the assistant can query us directly instead, it gets current, verified carrier facts. When its user is ready, it can request quotes on their behalf, which reaches a licensed human with the details already in place. The human advisor still runs the medical questions, the carrier comparison, and the negotiation.
The volume is still small. Most of our quote requests still arrive the ordinary way, typed by humans into a form on our website, and that will probably stay true for a while. We built the MCP because the researching phase of the process moved, and we would rather an assistant answering questions about disability insurance work from verified facts than from a compilation of guesses.
What should you take from this as a buyer?
If an assistant provides insurance numbers, ask it where they came from. Current assistants can produce both retrieved figures (a source you can check), and generated ones (a guess), while not distinguishing between the two. Anything quoted before underwriting is an estimate, no matter how precise it may appear.
And when a machine helps you compare policies, notice what it is comparing. The value in a disability contract is far more than just price. It sits in language, the own-occupation definition and the exclusions, and that is the part a machine is worst at weighing.
If your assistant speaks MCP, you can point it at mcp.seaworthy.io and see for yourself. The request it submits lands with a licensed advisor.
Frequently asked questions
Can an AI agent get you a disability insurance quote?
Will an AI agent be able to buy a disability insurance policy end to end?
Why can auto and home insurance be bought inside ChatGPT when disability insurance cannot?
What should you ask an AI assistant that gives you insurance numbers?
More from insights
- AI Sent Me Back to School at 43 Why AI sent a twenty-year disability insurance operator back to school for the CLU designation. AI made surface knowledge free and, in doing so, made verifiable human expertise the thing that now carries the weight.
- The Three-Minute Underwrite AI's quietest talent is removing the cost of a human reading your file: 1.2-second claim denials, three-minute quotes, and what that speed misses.
- AI Forced Us to Rebrand After fifteen years, Disability Insurance Agency became Seaworthy Insurance. The reason was AI: assistants recommend specialists they recognize as distinct entities, and a name that reads as the category gets folded into the summary instead of named as a choice.