Leads
B2B lead generation agent that finds companies and decision makers matching your ICP
Lead Researcher turns your ideal customer profile (ICP) into concrete search queries, verifies each prospect on its own website, enriches it and saves it as a lead with a 0–100 fit score and the evidence behind it.
- As a workflow node
agent.leadResearcher- Default model
- Claude Sonnet 5
- Economy model
- Claude Haiku 4.5
- Tools
- 6
- Report
- Lead list
What the agent does
For B2B lead generation the agent first breaks your profile into filters: industry and niche, region, company size, buying signals (runs ads, has an online store, is hiring, recently raised money) and the target job title. It then writes 6–10 varied queries in the market's language: directories and marketplaces, LinkedIn company pages, local listings, association member lists, award lists and job boards. Before searching it checks your existing leads, so companies already on your list are not suggested again.
For every promising prospect it opens the website to confirm the fit, enriches company data (size, location, LinkedIn, technologies) and, when email is among the preferred channels, looks up a business email. The fit score rests on five criteria: profile match, reachability, buying signals, geography and language, and timing. Prospects scoring 50 or more are saved the moment they are found, so nothing is lost if a run stops early.
Only contact details businesses publish themselves (website, LinkedIn company page, directories) are stored; no personal phone numbers are scraped from private profiles. Leads land on the Leads page, and records matching an existing email, phone or website are not duplicated. The usual next steps are Lead Qualifier and Outreach Agent.
Inputs / Outputs
Inputs
- Ideal customer profilerequired
- Make it measurable: industry, city, headcount and at least one signal. Example: “Dental clinics in Istanbul with 10–50 staff that are active on Instagram; no chains; decision maker is the clinic owner.”
- Region / country
- City, region or country to search in:
Turkey,Istanbul,DACH. Leave it empty to use the place in your profile or your project's market. - Target number of leads
- How many leads to find (default 20). It returns fewer when no more good matches exist; 20 a week, repeated regularly, is a good start.
- Preferred contact channels
- Choose from email, LinkedIn, WhatsApp and Instagram; the agent looks for contact details on these channels and suggests one per lead. Left empty, email and LinkedIn are used. Instagram is only for finding handles; OrqLabs doesn't send Instagram DMs.
Outputs
- leads
- Saved leads: company, decision maker and title, website, email, phone and LinkedIn when found, country and city, industry, size, a 0–100 fit score, a 1–2 sentence evidence-based reason, the suggested channel and the source URL.
- searchSummary
- Search summary: queries used, sources covered, how many candidates were reviewed and what was excluded and why.
Saves a report to the Reports page on every run: Lead list.
Tools
- Web search
web_search - Read a web page
fetch_url - Find an email
find_email - Company enrichment
enrich_company - Add a lead
create_lead - List leads
list_leads
Integrations it uses
Example tasks
- “Find dental clinics in Istanbul with 10–50 staff that are active on Instagram. Look for the owner or manager as the decision maker; skip clinic chains.”
- “List cosmetics brands in Turkey that run their own online shop. Put those with more than 10k Instagram followers first and find the founder's email.”
- “Find software companies with 20–100 staff in Ankara and Istanbul that could buy office coffee. Try to reach the office manager or HR manager.”
- “Find B2B SaaS companies that posted a marketing specialist job in the last 30 days; note the job ad as a growth signal in the reason.”
Tips for good results
- Make the profile measurable: industry, city, size and at least one signal (technology, job ad, social following, funding).
- Say who you want to reach (founder, marketing manager, clinic owner); otherwise you get generic contact addresses.
- Add exclusions: “no agencies”, “none of our existing customers”.
- Web search runs through Serper or Tavily; connect Hunter or Apollo for email finding and company enrichment.
- Use a saved agent: its memory keeps exhausted queries and rejected domains, so each week it focuses on new prospects.
Model advice
Default model
Claude Sonnet 5
Per 1M tokens: $2 input · $10 output
Economy model
Claude Haiku 4.5
Per 1M tokens: $1 input · $5 output
The default balanced model is a good balance for filtering many search results against your profile and writing evidence-based fit reasons. With a sharp profile, a narrow industry and a small target count, the economy model cuts the cost noticeably, though its reasons may be shallower. Credits for search and email-finding services (Serper, Tavily, Hunter, Apollo) come on top of the model cost.
Templates that use this agent
Frequently asked questions
Where does it find leads?
Web search, companies' own websites, directories and marketplaces, LinkedIn company pages and the Hunter or Apollo databases. Every lead record includes the source URL it was found on.
Which accounts do I need to connect?
Serper or Tavily for web search, and Hunter or Apollo for email finding and company enrichment. Saving leads needs no extra account.
Will it add the same company twice?
No. It checks your existing leads before searching, and records matching an existing email, phone or website are not duplicated. A saved agent also remembers the domains it rejected.
How does it handle personal data?
It only stores business contact details that companies publish themselves and never scrapes personal phone numbers from private profiles. Still mention the data source and an opt-out in your first messages.
Get started with OrqLabs
Build your first workflow today
Start on the free plan: connect your accounts, pick a template and stay in control with approval steps. No credit card required.