Claude + EZLynx:
what works today, and what doesn’t.

The gate: EZLynx Connect, not a developer portal

EZLynx’s own API page promises real-time data exchange, automated quoting and custom workflows, and its only call to action is a contact form. There is no public documentation, no published price and no sign-up for a key.

Access runs through EZLynx Connect, the vendor program EZLynx set up in 2019. Every vendor that receives consumer data goes through legal and security review, and EZLynx has said some vendors also charge the agency a license fee before an integration is switched on.

• Ask your EZLynx account manager for API (Web Services) access for your agency, or work with a vendor that already has it.
• Expect the approval, not the build, to be the slow part.
• Authentication is a user name, password and app secret that return a session token, not OAuth. Plan for a dedicated API user.

What EZLynx gives you, and what it doesn’t

Rating is what EZLynx is known for, and it keeps rating inside its own product: Instant Commercial Quoting and the comparative rater are features, not public endpoints. Applied’s AI announcements this year (the EZLynx roadmap’s “AI-powered AMS” and Epic Conductor on the Applied Epic side) are built into Applied’s products. None mentions MCP, Claude or bringing your own AI.

What the API does cover, going by a public reference collection that has since been taken down (confirm the current list with EZLynx):

• Read and write applicants (personal and commercial), contacts, prospects and opportunities.
• Read policies, including the policy’s ACORD XML and transaction history, and add a policy or rewrite.
• Read a completed quote by its id, but not start a rating. Outside quotes can be added and show up as manual quotes.
• Upload a document from a URL and read the document library.
• Webhooks for about thirty events (applicant, policy, opportunity, quote completed, document created, policy downloaded). EZLynx registers your URL during account setup, and each payload carries only an id you fetch back.
• A Zapier app with applicant, prospect, opportunity, document and note triggers. It is not open to every agency; your account manager enables it.

How you would actually connect Claude to EZLynx

With access in hand, the bridge is a small MCP server: a program that tells Claude which EZLynx actions exist and makes the API calls when Claude asks. Someone technical builds it once and keeps it running.

1. 01
   Get API access for your agency

   Through your EZLynx account manager or EZLynx Connect. Ask for a dedicated API user so the audit trail shows what the AI did.
2. 02
   Build the MCP server

   Wrap the endpoints you need as tools. The server holds the credentials, caches the session token and logs in again when it expires.
3. 03
   Choose where it runs

   As a remote server added to Claude as a custom connector (reachable from the internet, OAuth in front), or locally beside Claude Desktop so credentials never leave the office.
4. 04
   Put a person in front of every write

   Read tools can run freely. Creating a prospect or uploading a document asks for confirmation first.
5. 05
   Add a listener for events

   Claude does not receive webhooks. If “new applicant” should trigger work, a separate receiver catches the EZLynx webhook and calls the Claude API itself.

The limits that actually matter

• No rating through Claude. A person runs the rater; Claude can read the result afterward.
• Webhooks exist but EZLynx configures them for you, and each event has to be fetched back to see the record.
• Which EZLynx editions include API access, and what it costs, is not published.
• The only detailed public reference has been taken down, so confirm every endpoint with EZLynx before building on it.
• Session-token authentication is one shared login, so per-person permissions have to be built in your bridge.

Claude itself has one more limit on every AMS: it acts when someone asks, or on a scheduled task’s clock. It does not notice a new lead, a changed policy or a cancellation notice on its own. Anything that should happen because something changed in EZLynx needs a listener running outside Claude.

If you want EZLynx work that runs on its own

We are an engineering team that builds automation on top of the AMS an agency already runs. For EZLynx that means one of two things: we build and host the bridge above so your team can ask Claude about the book, or we skip the chat window and build the workflow itself, triggered by EZLynx events and checked by rules where a wrong number matters.

• A hosted Claude connector on your EZLynx data, with read tools for the whole team and writes that ask first.
• Webhook-driven work with no chat window: a new applicant gets its ACORD prepped, a downloaded policy gets checked against the quote.
• Carrier portal and rater entry from the EZLynx record, with quotes written back, for the steps the API leaves out.

The first build is live in 14 days for $2,499, charged only if your team is using it, then $2,499 a month flat. You keep the code. More on what we automate on top of EZLynx and our EZLynx integration engineering.

Trucking agency? Our Market Tracker connector already works in Claude today: it answers carrier questions (filings, X-dates, safety, ISS) from the federal record, so the AMS bridge only has to carry your own book.

Claude + EZLynx, answered.