Apollo

Routing layer for Apollo API work. WHAT/guardrails here; HOW in reference/endpoints.md (59 ops) and the Python client lib/apollo.py. Full per-endpoint OpenAPI is cached at reference/docs/apollo/raw/<slug>.md.

1) Auth & config

• Base URL: https://api.apollo.io/api/v1
• Auth: API key in header x-api-key (Apollo → Settings → API). Env APOLLO_API_KEY; never hardcode.
• Pagination: page + per_page; response pagination:{page,per_page,total_entries,total_pages} + a result array (people/contacts/accounts/organizations). Client _paginate follows it.

2) Credits & outreach guardrails (important)

• Search ≠ Enrich. search_people / search_organizations are net-new prospecting — they return no emails/phones and cost no credits. Use them to build shortlists.
• Enrichment costs credits. enrich_person (/people/match), bulk_enrich_people, enrich_organization reveal emails/phones and consume credits. reveal_personal_emails / reveal_phone_number flags gate paid reveals — default off, confirm before bulk.
• Sequences = real outreach. add_contacts_to_sequence enrolls people into live sending and activate_sequence starts it — approval-gated (pilot one, confirm, then batch).
• Pilot bulk enrichment/writes on 1 record first; show payload+response; get explicit go-ahead.

3) Operations covered by the client

People (search, enrich, bulk_enrich), organizations (search, enrich, bulk_enrich, get, job_postings), contacts (create, update, view, search, bulk_create), accounts (create, update, view, search), sequences (search, add_contacts, activate, deactivate), lookups (fields, email_accounts, users, lists). Anything else in the 59-op surface → generic request(method, path, …) + reference/endpoints.md.

4) Handoff with the GTM/recruiting layer

Apollo is a sourcing + enrichment engine. Canonical flow: search_people/search_organizations to build a net-new shortlist (free) → enrich_person only on the shortlist (credits) → write to your ATS (recruit-crm) or hand to a sequencer (instantly/sourcewhale). Enrich late, on the filtered set, never the whole funnel (over-provision then filter).

Field notes (production experience)

Field-shape note: these are vendor-native operational notes. The client returns the raw vendor JSON and uses the method names in this file — read field shapes accordingly (no normalized-wrapper / result.data. prefix).

Response shape (our direct client): people search returns top-level { total_entries, people, pagination } - read response["people"]. Company search is canonical at organizations (not accounts) - read response["organizations"]; rows carry name, primary_domain/domain, linkedin_url, industry.

• Free vs paid people search: mixed_people/api_search = free preview (obfuscated names/contact gaps, no credits) - use for cheap discovery/shortlisting. mixed_people/search = paid (full coverage). Start free, escalate to paid only when needed.
• Obfuscated last names: detect S****, K., -, N/A, masked strings from free search; do not feed them to email-finders/pattern generators. Bridge: free search -> people match by Apollo id -> use the resolved last_name. Flag last_name_obfuscated=true.
• Company targeting: keep include_similar_titles=true unless strict matching is asked. Prefer q_organization_domains_list/organization_ids over name-only keywords. For exact industry use organization_industry_tag_ids (taxonomy IDs from industry_tag_id/industry_tag_hash in org rows - Apollo doesn't publish the taxonomy in the spec), not q_organization_keyword_tags (keyword search, pulls off-industry).
• organization_num_employees_ranges is contract-sensitive - use the exact format Apollo expects; don't trust pagination.total_entries alone - pilot with low per_page first.

Callable surface — lib/apollo.py

Import: from lib.apollo import Apollo → instantiate Apollo() (reads key from env). Base: https://api.apollo.io/api/v1. Generic passthrough: request(method, path, *, params, json).

• activate_sequence(sequence_id: str) -> dict — POST /emailer_campaigns/{id}/approve — starts the sequence sending.
• add_contacts_to_sequence(sequence_id: str, contact_ids: list[str], email_account_id: str, **opts) -> dict — POST /emailer_campaigns/{id}/add_contact_ids — ENROLLS contacts into live outreach.
• bulk_create_contacts(contacts: list[dict], **opts) -> dict
• bulk_enrich_organizations(domains: list[str], **opts) -> dict — POST /organizations/bulk_enrich — up to 10 domains per call.
• bulk_enrich_people(details: list[dict], **opts) -> dict — POST /people/bulk_match — enrich up to 10 people per call. Consumes credits.
• create_account(payload: dict) -> dict
• create_contact(payload: dict) -> dict — POST /contacts — create a CRM contact in your Apollo.
• deactivate_sequence(sequence_id: str) -> dict
• enrich_organization(domain: str, **params) -> dict — GET /organizations/enrich — enrich one company by domain.
• enrich_person(payload: dict) -> dict — POST /people/match — enrich one person (email/phone). Consumes credits.
• get_organization(org_id: str) -> dict
• list_email_accounts(**params) -> Any — GET /email_accounts — sending mailboxes (need an id to enroll into sequences).
• list_fields(**params) -> Any
• list_lists(**params) -> Any — GET /labels — Apollo lists.
• list_users(**params) -> Any
• organization_job_postings(org_id: str, **params) -> Any
• search_accounts(filters: dict | None = None) -> Iterator[dict]
• search_contacts(filters: dict | None = None) -> Iterator[dict] — POST /contacts/search — search your Apollo contacts.
• search_organizations(filters: dict | None = None) -> Iterator[dict] — POST /mixed_companies/search — company/account search.
• search_people(filters: dict | None = None) -> Iterator[dict] — POST /mixed_people/api_search — prospect net-new people. No emails/phones, no credits.
• search_sequences(filters: dict | None = None) -> Iterator[dict] — POST /emailer_campaigns/search.
• update_account(account_id: str, payload: dict) -> dict
• update_contact(contact_id: str, payload: dict) -> dict — PATCH /contacts/{contact_id}.
• view_account(account_id: str) -> dict
• view_contact(contact_id: str) -> dict