Needle

Founders and growth teams who run outbound today often build lead lists manually using tools like Clay and LinkedIn, a process one user describes as the exact filter they run by hand. Finding the right ICP, personalizing messages with real context, and managing multi-step sequences across LinkedIn, Instagram and email takes significant manual effort. Teams also risk account restrictions from sending too aggressively, and without suppression logic they end up messaging existing customers, open deals or people colleagues already contacted. Doing this well at speed while keeping accounts healthy is described as hard to sustain manually.

Needle ships as an MCP server that connects directly to agents like Claude, ChatGPT, Codex, Cursor or Grok Bot, letting users ask their agent to find leads and launch campaigns in natural language. It searches LinkedIn, Instagram and Places to build ICP lead lists, drafts messages grounded in a lead's profile, conversation history, recent posts and comments, and runs multi-step, multi-channel sequences (e.g., LinkedIn message, wait, follow-up email). Built-in guardrails include daily send limits, human-like cadence, warm-up schedules for new senders, and automatic suppression of existing customers, open deals or recently contacted people. It also surfaces funnel analytics and insights (e.g., reply timing, messaging angle performance) to help refine campaigns over time.