Placeholder entry. No source material for agents exists yet, so this page is here to establish the shape of the catalog, not to describe something you can run. Its name, its run cadence, and where it stops for approval are invented. What is real: the prompts, the skills listed under Skills this agent would chain, and, where the page has one, the verbatim turn sequence. See the gaps list in data/catalog.yaml.

What it does

Chains prospecting and campaign tools into one task: build the list, draft a multi-touch sequence, and hold at review. The Prompt Library’s campaign use case is already written this way: a list-building prompt, then scheduling, then “show me the draft emails to review before anything sends”.

Skills this agent would chain

  1. Renewal X-date prospect list: Every company in your territory renewing in the next 60 to 120 days that isn’t already in your book, ranked and ready to work.
  2. Competitive displacement campaign: Target the accounts a named competitor holds and sequence them, with the competitor never named in an email.
  3. Vertical prospecting campaign: Turn an industry, geography, and size into a reviewed prospect shortlist, a multi-touch email sequence, and a scheduled send, with two confirmation gates.
  4. Prospect-to-account conversion: Turn the companies a prospecting skill found into CRM prospect accounts, populated and de-duplicated, in one confirmed batch.

Each of those is a real catalog entry you can run on its own today.

The turns this would automate

  1. Build a 3-touch email campaign to the self-funded prospects we just found.
  2. Set the campaign to morning sends, starting Monday.
  3. Show me the draft emails to review before anything sends.
  4. Rotate timeline and social-proof hooks across the touches.

The four Employee Benefits prompts for “Launch a prospecting campaign” in the Zywave MCP Prompt Library, verbatim and in source order. The Commercial and Personal Lines sets have the same four-step shape. “the prospects we just found” is the Prompt Library’s own wording, and it is what suggests this use case already chains from a discovery turn.

Today you type those yourself, one after another. What an agent would add is running them as one task and holding at the review step, and that part is the placeholder, not the turns.

What you get back

  • A prospect list and the campaign built from it
  • Draft emails held for your approval before any send
  • A schedule you confirm rather than one it sets for you

Example prompt

Add the agent to your AI client first, then paste this. Swap the bracketed bits, and the names and numbers, for your own.

Build a 3-touch email campaign to the self-funded prospects we just found.

Source: Verbatim from the Prompt Library (Employee Benefits · Launch a prospecting campaign). The agent around it is placeholder.

Open Claude Open ChatGPT

These buttons try to pre-fill the prompt for you. Open Claude launches the Claude Desktop app (it won't work if you only use Claude in a browser); Open ChatGPT pre-fills the web chat when the browser allows it. Neither is guaranteed, so both also copy the prompt to your clipboard as a fallback -- if the tool opens with an empty box, paste. Connect the Zywave MCP in that tool first, or the prompt has no Zywave data to work with.

The same shape, other lines of business

Employee Benefits

Show me the draft emails to review before anything sends.
Open Claude Open ChatGPT

Commercial (P&C)

Build a 4-touch email campaign to the trucking prospects we just found.
Open Claude Open ChatGPT

Source: Zywave MCP Prompt Library, unless the page says otherwise.

More prompts like this: Launch a prospecting campaign, 12 more from the Zywave MCP Prompt Library, filterable by line of business.

What it needs from you

A few details from you. You supply the specifics in the prompt, company, state, size, dates.

Details to fill in when you run it:

  • The segment to campaign to, and the send window you want

Sending is a write action. Nothing should leave your agency without an explicit approval step.

Guides exist for Claude, Microsoft Copilot, ChatGPT, and Google Gemini Enterprise. Custom integrations talk to the MCP server directly, so start from the MCP Apps reference for those.

Nothing from your clientComing soonRenewal prep agentplaceholderWorks a renewal list from end to end. Pulls the accounts renewing and assembles the notice packets.Manage your book of business Nothing from your clientComing soonBook hygiene agentplaceholderWalks your book looking for stale or missing data and proposes the fixes for you to approve.Manage your book of business

← Every agent in the Library