OpenClaw CRM: Complete Tutorial for 2026

Complete OpenClaw CRM tutorial for 2026. Learn how to set up DenchClaw on OpenClaw, configure objects, run AI queries, and build a production CRM.

Mark Rachapoom

March 16, 2026·7 min read

OpenClaw CRM: Complete Tutorial for 2026

OpenClaw is the AI agent framework that powers DenchClaw. Understanding the relationship between the two helps you get more out of both.

This tutorial covers DenchClaw as an OpenClaw application — how it's structured, what makes it tick, and how to configure it for production CRM use in 2026.

OpenClaw vs. DenchClaw: The Architecture#

OpenClaw is the underlying framework: an open-source AI agent runtime that handles sessions, channels (Telegram, Discord, WhatsApp, etc.), tool execution, multi-agent routing, and skills.

DenchClaw is the opinionated, batteries-included workspace layer built on OpenClaw. It adds: the CRM database (DuckDB + EAV schema), the App Builder, the gstack workflow, the browser agent, and all the pre-installed skills.

The analogy: OpenClaw is React. DenchClaw is Next.js.

When you run npx denchclaw, you're launching an OpenClaw gateway with the DenchClaw workspace profile — profile name dench, port 19001.

Installing for 2026#

The installation is unchanged from launch:

npx denchclaw

This bootstraps:

OpenClaw profile: dench
Web frontend: localhost:3100
DuckDB database: ~/.openclaw-dench/workspace/workspace.duckdb
Gateway port: 19001

For production deployments, use the VPS setup:

# On a fresh Ubuntu 22.04 VPS
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt-get install -y nodejs
npx denchclaw --host 0.0.0.0

Add Nginx as a reverse proxy with TLS, and you have a private CRM accessible anywhere.

The CRM Data Model#

DenchClaw's CRM uses an EAV (Entity-Attribute-Value) schema in DuckDB:

objects table        — defines object types (people, company, deal, etc.)
fields table         — defines fields/columns for each object
entries table        — one row per record
entry_fields table   — stores the actual field values
statuses table       — configures status options per object
documents table      — links markdown documents to entries

Pivot views (v_people, v_companies, v_deals) flatten this into readable, queryable tables:

-- Direct DuckDB query
SELECT "Full Name", "Email", "Status"
FROM v_people
WHERE "Status" = 'Lead'
ORDER BY "Full Name";

Every pivot view is auto-generated by the CRM skill when you create an object.

Creating Your Object Schema#

For a standard B2B SaaS CRM in 2026:

People:

Create a People object: Full Name, Email, Phone, Company (relation), Title, Status (Lead/Prospect/Customer/Churned), Last Contacted, Tags, Notes

Companies:

Create a Companies object: Company Name, Website, Industry, Size, ARR, Primary Contact (relation to People), Account Owner, Health Score

Deals:

Create a Deals object: Deal Name, Company (relation), Contact (relation), Value, Stage (Discovery/Proposal/Negotiation/Closed Won/Closed Lost), Close Date, Owner, Probability

Activities:

Create an Activities object: Type (Call/Email/Meeting), Date, Contact (relation to People), Deal (relation), Notes, Next Steps

Connecting Messaging Channels#

In 2026, DenchClaw supports more channels than ever. Configure your preferred ones:

Telegram (recommended for individuals):

Create a bot via @BotFather
Get the HTTP token
Add to DenchClaw settings: Gateway → Channels → Telegram → Add bot token

WhatsApp:

Scan the QR code in DenchClaw settings
Keep your phone connected to the internet

Discord (recommended for teams):

Create a Discord application and bot
Add to your server
Configure in DenchClaw with the bot token and guild ID

Slack:

Create a Slack app
Configure OAuth credentials
Add to your workspace

Each channel gives you full conversational access to your CRM.

Skills — The Extension Layer#

Skills are what make OpenClaw infinitely extensible. DenchClaw ships with these pre-installed:

crm, app-builder, gstack, browser, github, gog, himalaya, imsg, wacli, 
apple-notes, apple-reminders, things-mac, obsidian, here-now, nano-pdf,
video-frames, weather, coding-agent, peekaboo, stripe-best-practices

Install community skills from clawhub.ai:

Install the apollo-enrichment skill from clawhub

Install the clearbit-lookup skill

Each skill is a SKILL.md file that teaches the agent new capabilities — no compiled code, no npm packages.

Advanced Object Configuration#

Field Types#

DenchClaw in 2026 supports these field types:

text, richtext — plain and markdown text
number — numeric, configurable precision
email, phone, url — validated contact fields
date, datetime — with timezone support
enum — single-select dropdown
tags — multi-select
relation — links to another object
file — attached files stored in workspace
action — server-side executable button
user — maps to a workspace user

Action Fields in Depth#

Action fields are the most powerful and underused feature. An action field runs a server-side script when the button is clicked. The script:

Receives the full entry as environment variables
Can run any shell command, Python script, or API call
Streams output back to the UI as NDJSON
Can update entry fields directly

Example: an "Enrich from Clearbit" action field:

// action.js
const company = process.env.FIELD_COMPANY;
const response = await fetch(`https://company.clearbit.com/v1/domains/find?name=${company}`, {
  headers: { Authorization: `Bearer ${process.env.CLEARBIT_API_KEY}` }
});
const data = await response.json();
console.log(JSON.stringify({ type: 'update', fields: { Industry: data.category?.industry, Size: data.metrics?.employees } }));

Custom Views#

Views are configured in .object.yaml:

views:
  - name: Hot Leads
    filters:
      - field: Status
        operator: equals
        value: Lead
      - field: Last Contacted
        operator: older_than
        days: 14
    sort:
      - field: Company
        direction: asc
    default_columns: [Full Name, Email, Company, Status, Last Contacted]

Running the Agent#

The DenchClaw agent responds to natural language and executes structured operations:

Natural language queries:

Show me all deals over $50k closing this quarter

Structured operations:

Add Sarah Chen from Stripe to People, email sarah@stripe.com, status Lead

Complex analysis:

Which companies in my database have the highest deal velocity? 
Calculate average days from Lead to Closed Won grouped by company industry.

Background agents:

In the background, enrich all new contacts added in the last week

Production Considerations#

For teams running DenchClaw in production in 2026:

Backup: Add a cron job to copy workspace.duckdb to S3 or a backup location daily.

Monitoring: The OpenClaw gateway logs to ~/.openclaw-dench/workspace/logs/. Monitor for errors.

Updates: npx denchclaw always pulls the latest. For pinned versions: npx denchclaw@1.2.3.

Performance: For databases over 500k entries, add DuckDB indexes on frequently queried fields:

CREATE INDEX idx_people_email ON entry_fields(field_id, value) WHERE ...

For more on DenchClaw's full feature set, see what is DenchClaw. For a faster start, see the zero to CRM tutorial.

Frequently Asked Questions#

What's the difference between OpenClaw and DenchClaw?#

OpenClaw is the open-source AI agent framework (like Express.js). DenchClaw is the pre-built workspace application on top of OpenClaw (like a full web app built with Express). You can build your own OpenClaw apps, but DenchClaw gives you everything pre-configured.