A rep needs ten VPs of Sales at fintech companies under 500 people. Right now that means opening Lusha, setting five filters, exporting a list, and copying it somewhere useful. Every search starts from zero. This play replaces the filters with a sentence. Describe who you need in plain English, and an AI agent turns that into a Lusha search — then hands back verified contacts, ready to use.
I detailed how you can build this out below but If you need more information, you can find the full workflow tutorial in Lusha’s campus, other Lusha x n8n workflows or use the n8n template here.
The problem
Search filters are precise, but precise isn’t fast. A rep mid-conversation in Slack doesn’t want to open a new tab, remember filter names, and build a query. They want to ask for what they need and get it back in seconds.
What you’ll get
A working chat interface — Slack, Telegram, or n8n’s own chat window — where typing a request like “find me 10 VPs of RevOps at SaaS companies in the US” returns a list of verified contacts: name, title, company, and whether email and phone are available to reveal.
How to build it
01 Install the Lusha node and add your API credentials.
02 Add your model provider credentials — Claude or GPT-4o — to the AI Agent node.
03 Connect a chat trigger. Slack or Telegram both work the same way.
04 Set the agent’s system prompt: extract job title, seniority, company size, industry, and location from the request, then call the Lusha search tool.
05 Tell the agent to ask a follow-up question if the request is too vague, instead of guessing.
06 Test it: “Find me 5 VPs of Marketing at mid-market SaaS companies.”
A note on credits: the agent returns names, titles, and companies by default — not revealed emails or phones. Revealing contact details is a separate step, so credit spend stays a decision, not a side effect.
FAQ
-
Does this reveal emails and phones automatically?
No. It returns contact identification with availability flags. Revealing is a separate, deliberate action.
-
Slack or Telegram — does it matter which?
No. Swap the trigger node, the rest of the build stays the same.
-
What if the request doesn’t map to a clear filter?
The agent asks a clarifying question instead of running a bad search.
-
Do I need the MCP server or the community node?
Either works. The community node is the simpler path if this is your first agent build with Lusha.
Happy to answer any comments or questions in the comments below!