Inbound Lead Enrichment

Inbound Lead Enrichment

The enrichment arm feeding inbound-lead-qualification and inbound-lead-triage. Every step has a keyless web-research degrade; paid tools (Apollo, dropcontact, CRM) only deepen it. Synthesis is the agent's job — the scripts do deterministic fetch/lookup only.

When to use

• "Enrich these leads." / "Fill in the missing data on these inbound leads."
• When qualification flags leads insufficient_data, or triage finds missing fields.

How to run

Step 1 — assess gaps (you, the agent)

Inventory what's missing per lead (company profile, role/seniority, LinkedIn URL, email).

Step 2 — company + person research (keyless)

python3 ${SKILL_DIR}/scripts/web_research.py \
  --leads ${WORKSPACE}/leads.csv --fetch-pages \
  --output ${WORKSPACE}/research.json

Returns per-lead search hits + extracted company-page text + LinkedIn candidate URLs. You read these and synthesize the company profile, role/seniority, and LinkedIn URL. When APOLLO_API_KEY / DROPCONTACT_API_KEY are set, prefer those structured sources (via apollo-lead-finder / a dropcontact call) and use this to fill gaps. Verify any found email with an email-verify skill.

Step 3 — stakeholder discovery (buying committee)

Invoke company-contact-finder per company to find the rest of the committee; dedup against the original lead by LinkedIn URL/email.

Step 4 — relationship check (CRM)

python3 ${SKILL_DIR}/scripts/crm_lookup.py \
  --leads ${WORKSPACE}/leads.csv \
  --crm-csv ${WORKSPACE}/customers.csv \
  --output ${WORKSPACE}/relationships.json

Uses HubSpot search when HUBSPOT_API_KEY is set; otherwise matches a CRM/customer CSV export by email/company. Flags existing_relationship / company_overlap.

Step 5 — compile (you, the agent)

Merge research + stakeholders + relationship flags into enriched records. Flag rather than invent — if a field can't be found, mark it unknown, never fabricate.

Outputs

• research.json — [{company, person, company_hits, person_hits, linkedin_candidates, company_page_text}].
• relationships.json — [{name, email, company, existing_relationship, company_overlap, source}].
• Your compiled enriched records (company profile, role/seniority, stakeholders, flags).

Credentials / env

• Required: none — every step has a keyless serp/CSV degrade.
• Optional: APOLLO_API_KEY (primary company/person enrichment); DROPCONTACT_API_KEY
  • an email-verify key (email finding/verification); a CRM key (HUBSPOT_API_KEY / SALESFORCE_*) for the relationship check; PHANTOMBUSTER_API_KEY + cookie (LinkedIn profile depth).

Notes & edge cases

• Tool-agnostic: prefer the configured primary/secondary tool per category; fall through to web research when a paid tool is absent.
• Stakeholder discovery must not duplicate the original lead — dedup by LinkedIn URL/email.
• Flag rather than invent: unknown fields stay unknown.
• The keyless serp endpoint rate-limits; in production this maps to the Robomotion serp Search + Extract Content nodes with a proxy.