Pancake CRM Records CRUD API — Managing lead / account / contact

TL;DR: Pancake CRM exposes 4 Records endpoints for the 3 table types lead / account / contact: GET /workspaces/{ws}/{table}/records (list with JSON filter + cursor pagination), POST /workspaces/{ws}/{table}/records (upsert — id present = update, id omitted = create; there is NO separate create/update endpoint), DELETE /workspaces/{ws}/{table}/records?record_ids[]=... (bulk delete), and GET /workspaces/{ws}/record/{id} (single — different URL pattern from list). The filter supports 9 operators and AND/OR logic. Watch out for the special 422 LEAD_CANNOT_UPDATE_FROM_CANCEL error — you can't update a lead that has been canceled/converted/merged.

Once you've set up the Pancake CRM API + api_key, the most-used endpoint group is Records — where leads/contacts/accounts for the workspace live. This post covers all 4 CRUD endpoints, the full filter JSON syntax, and the common pitfalls you'll hit when integrating.

1. The 3 record types — which to use when?

Pancake CRM follows the standard industry CRM data model, splitting 3 entity types with different purposes:

Real-world scenarios

• B2C shop (cosmetics, fashion): primarily uses lead (for ads) + contact (paying customers). account is not needed.
• B2B agency (marketing, dev studios): uses account (the client company) + contact (people inside the company). One account, many contacts.
• Multi-channel SaaS: uses all three — lead from FB ads → contact on demo call → account when onboarding the org.

2. Setup recap

All endpoints below need:

• api_key — query param (see CRM intro section 4)
• workspace_id — path param (see section 3)
• table_name — path param, valid values: lead / account / contact (INVALID: customer, user, ...)

3. GET /records — list with cursor pagination

GET https://crm.pancake.vn/api/workspaces/{workspace_id}/{table_name}/records?api_key={key}

cURL example — list all leads

curl -G "https://crm.pancake.vn/api/workspaces/$WORKSPACE_ID/lead/records" \
  --data-urlencode "api_key=$API_KEY" \
  --data-urlencode "view_id=all"

Response shape

{
  "cursor": "eyJwYWdlIjoyfQ==",
  "entries": [
    {
      "id": "a1b2c3d4-...",
      "name": "John Doe",
      "phone_number": "+1234567890",
      "email": "john@example.com",
      "birthday": "1990-05-12T00:00:00Z",
      "full_address": "126 Hang Trong, Hoan Kiem Ward, Hanoi, Vietnam",
      "pancake_tags": ["104943019063372_11"],
      "source": [-11],
      "owner": ["user_uuid_1"],
      "created_on": "2026-06-20T10:30:00Z",
      "modified_on": "2026-06-22T14:15:00Z",
      "interested_product": ["Serum HA", "Cream B5"]
    }
  ]
}

Cursor pagination

Pancake CRM Records uses cursor-based pagination (same pattern as the chat list conversations endpoint):

• First request: do NOT pass cursor → response returns the next-page cursor
• Subsequent requests: pass the cursor from the previous response → get the next page
• End of data: response returns entries: [] or cursor: null/missing → stop the loop

# First request
curl -G "https://crm.pancake.vn/api/workspaces/$WS/lead/records" \
  --data-urlencode "api_key=$KEY"
# → {"cursor": "abc123...", "entries": [...]}

# Next request
curl -G "https://crm.pancake.vn/api/workspaces/$WS/lead/records" \
  --data-urlencode "api_key=$KEY" \
  --data-urlencode "cursor=abc123..."

4. Filter — JSON query with 9 operators

filter is the most powerful parameter of the list endpoint — JSON-encoded, supporting AND/OR logic + 9 operators.

Base structure

{
  "search_all": "0972273341",
  "is_group": false,
  "fields": [
    {
      "field_name": "phone_number",
      "type": "$having_value",
      "value": null,
      "is_filter_exclude": false
    }
  ]
}

9 operators

Example 1 — Find leads that have a phone

FILTER='{"fields":[{"field_name":"phone_number","type":"$having_value"}]}'

curl -G "https://crm.pancake.vn/api/workspaces/$WS/lead/records" \
  --data-urlencode "api_key=$KEY" \
  --data-urlencode "filter=$FILTER"

Example 2 — Leads with phone OR not yet contacted (OR instead of AND)

FILTER='{
  "is_group": true,
  "fields": [
    {"field_name":"phone_number","type":"$having_value"},
    {"field_name":"last_contact","type":"$having_no_value"}
  ]
}'

Example 3 — Leads from FB ad sources (source ID 10 or 20)

FILTER='{"fields":[{"field_name":"source","type":"$select_multi","value":[10,20]}]}'

Example 4 — Duplicate leads by phone (find duplicate phone numbers)

FILTER='{"fields":[{"field_name":"phone_number","type":"$duplicate"}]}'

Example 5 — Find leads NOT from FB (NOT)

FILTER='{
  "fields": [
    {
      "field_name": "source",
      "type": "$select_multi",
      "value": [10, 20],
      "is_filter_exclude": true
    }
  ]
}'

5. view_id — pre-defined dashboard views

In the CRM dashboard, users can save a view with a combination of filter + sort + display columns (e.g. "Hot leads this week", "Contacts not followed up in 7 days"). Each view has a UUID.

# Get the "all" view — includes every record with no filter
curl -G "https://crm.pancake.vn/api/workspaces/$WS/lead/records" \
  --data-urlencode "api_key=$KEY" \
  --data-urlencode "view_id=all"

# Get records using a saved view UUID from the dashboard
curl -G "https://crm.pancake.vn/api/workspaces/$WS/lead/records" \
  --data-urlencode "api_key=$KEY" \
  --data-urlencode "view_id=550e8400-e29b-41d4-a716-446655440000"

Use case: the sales team already set up a view "Hot Lead — has phone + not closed in 14 days" in the dashboard; your BI dashboard code just passes that view_id → no need to re-implement the filter logic, and stays in sync with the UI.

6. POST /records — Upsert pattern

POST https://crm.pancake.vn/api/workspaces/{workspace_id}/{table_name}/records?api_key={key}
Content-Type: application/json

Request body — Record schema

{
  "id": "a1b2c3d4-...",
  "name": "John Doe",
  "phone_number": "+1234567890",
  "email": "john@example.com",
  "birthday": "1990-05-12",
  "full_address": "126 Hang Trong, Hoan Kiem Ward, Hanoi, Vietnam",
  "pancake_tags": ["tag_id_1", "tag_id_2"],
  "source": [10],
  "owner": ["user_uuid"],
  "interested_product": ["Serum HA"]
}

Create — omit id

curl -X POST "https://crm.pancake.vn/api/workspaces/$WS/lead/records?api_key=$KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "John Doe",
    "phone_number": "+1234567890",
    "email": "john@example.com",
    "source": [10]
  }'

Response:

{
  "success": true,
  "data": {
    "id": "a1b2c3d4-newly-generated",
    "name": "John Doe",
    "phone_number": "+1234567890",
    "...": "..."
  }
}

Update — include id

curl -X POST "https://crm.pancake.vn/api/workspaces/$WS/lead/records?api_key=$KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "id": "a1b2c3d4-...",
    "name": "John Doe (verified)",
    "phone_number": "+1987654321"
  }'

Other errors

• 403 Forbidden — your api_key lacks permission for this table. Check the role + workspace ownership.

7. DELETE — bulk delete

DELETE https://crm.pancake.vn/api/workspaces/{workspace_id}/{table_name}/records?api_key={key}&record_ids[]={id1}&record_ids[]={id2}

record_ids[] is passed as a repeated query string (NOT a body):

curl -X DELETE "https://crm.pancake.vn/api/workspaces/$WS/lead/records?api_key=$KEY" \
  --data-urlencode "record_ids[]=uuid1" \
  --data-urlencode "record_ids[]=uuid2" \
  --data-urlencode "record_ids[]=uuid3" \
  -G

Response:

{
  "success": true,
  "fallback": "success_only"
}

8. GET /record/{id} — single record

The URL pattern is different from list (singular record/ instead of records), and table_name is a query param:

GET https://crm.pancake.vn/api/workspaces/{workspace_id}/record/{record_id}?api_key={key}&table_name={table}

curl -G "https://crm.pancake.vn/api/workspaces/$WS/record/$RECORD_ID" \
  --data-urlencode "api_key=$KEY" \
  --data-urlencode "table_name=lead"

Response:

{
  "success": true,
  "data": {
    "id": "a1b2c3d4-...",
    "name": "John Doe",
    "...": "..."
  }
}

Use case: when clicking a row in your internal dashboard → load full details via this endpoint (faster than filtering in the list).

9. Common pitfalls

10. FAQ

Q: Is there a bulk insert endpoint (POST many records at once)?

A: No. Pancake CRM only supports POSTing one record at a time. Bulk import = loop POST + throttle on the client side. Recommended pattern: sleep 100-200ms between requests, exponential backoff on 429/5xx.

Q: Is there built-in 2-way sync with HubSpot/Salesforce?

A: No official connector. Build it yourself via: HubSpot/Salesforce webhook → Pancake POST upsert + Pancake webhook → HubSpot/Salesforce API. See the 3 ways to extract Pancake data post for similar integration patterns (Pancake chat → CRM).

Q: Do custom fields (user-defined fields in the dashboard) appear in the response?

A: Yes — custom fields appear in the response with the names you set in the dashboard. The field schema in the docs only lists standard fields; for custom fields, GET a sample record to discover the actual schema for your workspace.

Q: What's the rate limit?

A: The official docs don't state a number. Recommended conservative throttle: 5-10 req/sec/workspace, with exponential backoff on 429. Test with production-like volume before going live.

Q: What's the difference between record (singular) and records (plural) in the URL?

A: The URL convention isn't consistent:

• /records (plural) = list / upsert / bulk delete
• /record/{id} (singular) = get single

Easy to confuse. When writing a wrapper class, define matching method names clearly.

11. Summary

• 4 Records endpoints: GET /records (list), POST /records (upsert), DELETE /records (bulk), GET /record/{id} (single — different URL)
• 3 table types: lead / account / contact — customer and other names are invalid
• Upsert pattern: id present = update, id absent = create — no separate create/update endpoints
• Powerful JSON filter with 9 operators + AND/OR + NOT — URL-encode when passing as a query
• view_id: use a dashboard view UUID to keep filter logic in sync with the UI
• Special error 422 LEAD_CANNOT_UPDATE_FROM_CANCEL — closed leads can't be updated, not even by admins
• Cursor pagination — loop until entries is empty; no total count

Related posts

Pancake CRM cluster

• Introducing the Pancake CRM API — pillar (auth, workspace, endpoint overview)
• This post — Records CRUD (you are here)
• Roadmap: Orders API, Tickets API, CRM Webhook, Pancake chat ↔ CRM integration

Pancake chat cluster (sister)

• Pancake Webhook & API Send Message Guide
• Get Pancake Conversations List API — cursor pagination pattern like Records
• 3 Ways to Extract Pancake Data into Your CRM — integration pattern reference

Official documentation

• Pancake CRM Developer Docs — OpenAPI spec

Last updated: 2026-06-23. API version: Pancake CRM 2.0.0. Check the OpenAPI spec to verify any changes.