Public connector SDK for JomWork — the managed AI workspace and company brain built for Malaysian businesses. Demonstrates how external tools format and submit payloads to a JomWork workspace for document ingestion, knowledge querying, and autonomous agent task submission.
JomWork indexes your company documents — Google Drive, Confluence, Notion, SharePoint, and 40+ more connectors — into a private RAG knowledge engine. Your team queries it in plain language (English or Bahasa Malaysia) and gets answers grounded in your actual documents, with source citations. Autonomous agents can act across connected platforms based on that knowledge.
Deployed on Malaysian cloud infrastructure with full data residency. Your data never leaves your environment, never touches another client's system, and is never used to train AI models.
jomwork_connector.py— the connector client: document ingestion, knowledge querying, agent task submission, and connector health checksexamples.py— eight worked examples mapping to real JomWork use cases: legal precedent indexing, HR policy Q&A in BM, multi-turn onboarding sessions, public FAQ ingestion, autonomous vendor contract renewal, logistics field access, clinic dual-brain deployment, and connector health monitoring
This SDK covers the ingestion and query boundary only. It does not describe JomWork's internal RAG pipeline, vector index, or agent runtime.
pip install httpxNo other dependencies. Python 3.9+.
import os
from jomwork_connector import JomWorkConnector
client = JomWorkConnector(
workspace_url=os.environ["JOMWORK_WORKSPACE_URL"],
api_key=os.environ["JOMWORK_API_KEY"],
)
# Ingest a document
client.ingest_document(
content="Annual leave entitlement for confirmed staff is 14 days per year.",
metadata={
"title": "HR Leave Policy 2026",
"source": "confluence",
"department": "HR",
"doc_type": "policy",
},
brain="internal",
)
# Query the brain
answer = client.query(
question="What is the annual leave entitlement for confirmed staff?",
brain="internal",
department="HR",
)
print(answer["answer"]) # grounded answer from your documents
print(answer["sources"]) # cited sources
# Submit an autonomous agent task
task = client.submit_agent_task(
instruction="Summarise all vendor contracts expiring Q4 2026 and draft renewal emails.",
connectors=["google_drive", "gmail"],
)| Surface | Method | Description |
|---|---|---|
| Text ingestion | ingest_document() |
Push plain text or markdown into the brain |
| File ingestion | ingest_file() |
Upload PDF, DOCX, XLSX, CSV, PPTX, TXT |
| URL ingestion | ingest_url() |
Fetch and index a web page or intranet URL |
| Knowledge query | query() |
Natural-language Q&A with source citations |
| Agent task | submit_agent_task() |
Multi-step autonomous task across connectors |
| Task status | get_task_status() |
Poll agent task result |
| Connector list | list_connectors() |
Inspect configured connector health |
JomWork runs two isolated brains per deployment:
- Internal Brain — authenticated staff access only. Full knowledge index. SSO/SAML/RBAC-protected.
- Public Brain — website chatbot. Only content explicitly ingested with
brain="public". Internal documents are never exposed.
The brain parameter on every ingestion and query call is the boundary. Content ingested to "internal" is never reachable from "public" queries.
Google Drive · Google Docs · Google Sheets · Confluence · Notion · SharePoint · Jira · Trello · Asana · Slack · Microsoft Teams · Gmail · Outlook · Dropbox · OneDrive · Box · HubSpot · Salesforce · Zendesk · Freshdesk · GitHub · GitLab · and more.
All connectors are configured from the JomWork admin console — no API credentials to manage in your application code.
JomWork is deployed as a managed service on Malaysian cloud infrastructure (AWS MY, Azure MY). On-premise and private cloud deployments available on Enterprise plan. No mandatory SaaS dependency.
- PDPA-aligned by design
- Data residency fully configurable
- Full audit log of all ingestion and query events
- Role and group permissions per department