Find Leads for Any ICP

AI lead research for any ICP across LinkedIn, the open web, and optional social platforms. Finds, ranks and exports leads with contact routes. Pay per pack, no monthly fee.

AI-ranked leads for any ICP, exported to CSV

Created by Emil Hjorth • Version 6 • 67 steps

Use this app

What you get

  • AI lead research for any ICP across LinkedIn, the open web, and optional social platforms. Finds, ranks and exports leads with contact routes. Pay per pack, no monthly fee.

Integrations

  • apify
  • openai
  • outreach-lead-files
  • Find Leads

How it works

  • Import demo video as same-flow publish media
  • Coerce string form-inputs to numbers
  • Start company-website crawl
  • Fetch + persist crawl dataset
  • Extract + truncate crawl text
  • Build company-brief LLM prompt
  • Generate company brief
  • Parse + validate brief JSON
  • Persist company brief blob
  • Detect whether user supplied keywords
  • Build keyword-inference LLM prompt
  • Auto-infer keywords from campaign goal + company brief
  • Parse inferred keywords JSON
  • Resolve final keywords (user > inferred)
  • Prepare per-platform Apify actor inputs
  • Serialize IG actor input to JSON for helper
  • Coerce helper output to inline-shape
  • Normalize YouTube dataset
  • Normalize TikTok dataset
  • Normalize LinkedIn dataset
  • Normalize X dataset
  • Normalize web search dataset
  • Coerce filter inputs to numbers
  • Gather + tag candidates (filter follower range + engagement, extract emails)
  • Split candidates into LLM-safe batches
  • Build LLM filter prompt (batch 1)
  • Build LLM filter prompt (batch 2)
  • Build LLM filter prompt (batch 3)
  • Build LLM filter prompt (batch 4)
  • Build LLM filter prompt (batch 5)
  • Persist LLM batch inputs
  • Filter candidates with OpenAI (batch 1)
  • Filter candidates with OpenAI (batch 2)
  • Filter candidates with OpenAI (batch 3)
  • Filter candidates with OpenAI (batch 4)
  • Filter candidates with OpenAI (batch 5)
  • Merge LLM decisions onto candidates
  • Compute final run status
  • Prepare candidate rows for table upsert
  • Upsert candidates table
  • Prepare outreach-ready lead rows
  • Upsert outreach-ready leads table
  • Prepare human-readable lead review rows
  • Persist human-readable lead review table
  • Build markdown report
  • Persist markdown report blob
  • Build JSON and CSV lead exports
  • Persist outreach leads JSON export
  • Persist outreach leads CSV export
  • Prepare run row
  • Upsert runs table

Find leads for any ICP, ranked by AI and ready for outreach

Lead Research is a pay-per-pack AI prospecting tool. You describe your ICP (ideal customer profile), it searches LinkedIn and the open web, ranks the candidates with AI, and exports a CSV with contact routes. No monthly subscription. Packs do not expire.

Built for sales teams, agencies, recruiters, PR leads, and founders who need a clean list of real people matching a target ICP, fast.

How does Lead Research work?

You give it three things: your company name, an ICP description, and a target lead count. Optional inputs: your company URL (so the AI ranker can read your brand context), which platforms to search, and your preferred contact channel.

The flow runs in roughly 90 seconds to 4 minutes per pack run:

  • Crawls your company website to build a brand brief
  • Infers search keywords from your ICP description
  • Runs parallel searches across LinkedIn and the open web (and optional Instagram, TikTok, YouTube, X)
  • Pulls candidate profiles and bios via Apify scrapers
  • Ranks each candidate with GPT-class AI against your ICP and brand brief, classifies as keep/review/reject with reasoning
  • Resolves contact routes (email when discoverable, LinkedIn DM, manual review)
  • Exports CSV, JSON, and a markdown report you can paste into Notion, HubSpot, Pipedrive, Salesforce, Clay, Instantly, or any CRM

How much does Lead Research cost?

Three usage packs. Pay once, run when you need it. No subscription.

  • Starter: $19 for 25 runs ($0.76 per run, recommended starting point)
  • Growth: $49 for 75 runs ($0.65 per run)
  • Agency: $129 for 250 runs ($0.52 per run)

Each pack includes a 1-run free trial with no credit card so you can verify the output quality before paying.

For context, comparable tools charge monthly:

| Tool | Monthly cost | What you get | |---|---|---| | Apollo.io | $99/mo (Professional) | 12,000 credits/yr, contact data + sequences | | Clay | $149/mo (Starter) | 5,000 credits/mo, AI research workflows | | Hunter | $49/mo (Starter) | 500 monthly searches, email finder | | Lead Research | $0.52–$0.76 per run, packs do not expire | Search + AI ranking + contact routes + CSV export |

A full Lead Research pack costs less than a single month of any of these tools. Runs do not expire when the month does.

Who uses Lead Research?

Sales teams, agencies, recruiters, PR leads, founders, and growth operators. Anyone who needs a CSV of qualified decision-makers without paying $99 to $499 a month for a dashboard.

Specific use cases buyers have run successfully:

  • B2B sales prospecting: "VPs of Engineering at 50-500 person SaaS companies in Europe and North America"
  • Account research: "Heads of Marketing at e-commerce brands doing $5M–$50M ARR"
  • Recruiting: "Senior backend engineers in fintech with Rust experience"
  • PR outreach: "Tech journalists covering AI tools at TechCrunch, The Information, and Stratechery"
  • Partnership research: "DevRel leads at developer-tool startups in Europe"
  • Influencer / creator outreach: "Independent biotech researchers active on Twitter/X" (or use the sibling Automatic Influencer Research for creator-specific search)
  • Investor research: "Series A B2B SaaS founders who raised in the last 12 months"

What do I need to provide?

Required: your company name, an ICP description in plain English.

Optional but recommended: your company URL (the AI uses it to build a brand brief for ranking), the platforms you want to search, and your preferred contact channel.

The ICP description is the only input that meaningfully changes output quality. Specific descriptions return specific leads. Vague descriptions return vague leads.

Strong ICP example: "Heads of Customer Success at vertical SaaS companies serving construction or manufacturing, 100-500 employees, North America"

Weak ICP example: "B2B leads in tech"

How does the AI ranking work?

Every candidate runs through a classification step that returns one of three decisions: keep, review, or reject. Each decision has reasoning attached.

The ranker uses:

  • Your ICP description (primary signal)
  • The brand brief from your company URL (if provided)
  • The candidate's bio, posts, and evidence URLs
  • Profile details from the platform of origin

Decisions are honest. A typical run returns roughly 15-25% keep, 20-30% review, and 50-60% reject. The reject list is preserved with reasoning, so you can audit and override.

How do I run it?

Three ways:

  • UI: go to https://breyta.ai/apps/lead-research, install, fill the form, run.
  • CLI: breyta flows run lead-research --input '{"icp-description": "...", "company-name": "..."}'. Good for scripted batch jobs.
  • MCP: Lead Research is exposed as the MCP tool find_leads. Any MCP-capable agent (Claude, ChatGPT, Cursor, your own agent) can call it directly. Useful for content-calendar agents, RevOps automations, and "find me 10 prospects for this campaign" agentic flows.

The MCP path takes the same input as the UI form. Required: icp-description. Everything else has defaults.

What you get back from each run

  • Ranked lead list with keep/review/reject classification and reasoning
  • Contact routes per lead: email (when discoverable), LinkedIn DM, Instagram/TikTok/YouTube/X DM, or manual review
  • CSV export: drop into HubSpot, Salesforce, Pipedrive, Clay, Apollo, Instantly, Lemlist, Smartlead
  • JSON export: for scripted automation or pushing into your own pipeline
  • Markdown report: paste into Notion, Linear, Slack, or HubSpot notes for the rest of the team

Files are saved to a private blob storage folder you control. Runs are also stored as Breyta tables you can query later.

Sibling flows in the same suite

Lead Research is one of four B2B outbound flows on Breyta. Each handles one part of the prospecting-to-reply pipeline:

Install one or all four. Same pay-per-pack pricing model on each. No monthly subscription anywhere.

Limits and failure cases

Lead Research relies on what each platform makes publicly available. LinkedIn rate-limits aggressively. Some profile fields will be missing. Email enrichment depends on what is discoverable on the open web at search time. Private profiles, blocked pages, and accounts with no public bio cannot be researched.

If a search returns zero candidates for a given ICP and keyword set, the flow returns an empty leads list with the source-status table showing which platforms returned nothing. You are not charged for an empty run.

The AI ranker is honest, not optimistic. If your ICP description is vague, the keep rate will be low. Rewrite the ICP and run again.

FAQ

What does the Find Leads for Any ICP app do?

This agent crawls the open web and social platforms to find leads that match your specific Ideal Customer Profile. It identifies decision makers on LinkedIn, X, and other social sites, then ranks them based on how well they fit your criteria.

How does the AI lead research process work?

The workflow starts by analysing your company website to understand your goal. It then uses Apify actors and OpenAI to search LinkedIn, TikTok, YouTube, and the wider web to find and verify contact details for relevant prospects.

Which platforms and services does this app connect with?

This app integrates with Apify for data collection and OpenAI for lead ranking and analysis. Once the research is finished, you can export your leads as CSV or JSON files ready for your outreach tools.

How do I set up and pay for lead generation?

You don't need a monthly subscription to use this tool because it's a pay-per-pack service. You simply input your campaign goals or keywords, and the agent handles the research, filtering, and report generation for you.