OpenClaw for Founders: The Complete Getting-Started Guide
OpenClaw is the AI agent framework powering DenchClaw. This guide explains what OpenClaw is, why founders should care, and how DenchClaw makes it accessible.
OpenClaw for Founders: The Complete Getting-Started Guide
If you've installed DenchClaw, you've already installed OpenClaw — you just didn't have to think about it. DenchClaw is the opinionated, founder-friendly layer on top of OpenClaw, the underlying AI agent framework.
But I want to explain OpenClaw to founders directly, because understanding what it is helps you understand what DenchClaw can really do — and what you can build on top of it.
What Is OpenClaw?#
OpenClaw is an open-source AI agent framework. Think of it as the operating system for AI agents on your machine. It handles:
- Sessions: Isolated conversation contexts with their own history and state
- Channels: Connecting the agent to messaging platforms (Telegram, WhatsApp, Discord, etc.)
- Tools: The agent's abilities — read/write files, execute code, browse the web, query databases
- Routing: Multi-agent orchestration (main agent spawning subagents)
- Memory: Session history and persistent file-based memory
- Events: Cron jobs, webhooks, heartbeat polls
OpenClaw was created by Peter Steinberger, formerly known for creating PSPDFKit and other developer tools. It's MIT licensed, actively developed, and growing a community.
The analogy that resonates: OpenClaw is to DenchClaw as React is to Next.js. OpenClaw is the powerful primitive. DenchClaw is the opinionated, batteries-included framework that makes it immediately useful.
Why Founders Should Care About OpenClaw#
As a founder using DenchClaw, you're running OpenClaw under the hood. Understanding this unlocks several things:
You can extend beyond CRM: OpenClaw supports any kind of agent workflow. DenchClaw is configured as a CRM agent, but you can add capabilities — build a custom investor relations agent, a hiring coordinator agent, a customer success agent — all running on the same OpenClaw runtime.
You can connect any messaging channel: OpenClaw supports 20+ channels out of the box. If your team is on Slack, connect it. If your investors are on Signal, connect it. If your customers are on LINE (common in Asia), connect it.
You can automate any workflow: OpenClaw's cron and event system can trigger agent actions on any schedule. Nightly data enrichment, weekly investor updates, daily standup summaries — all configurable.
You can spawn specialized agents: OpenClaw supports multi-agent architectures. Your main DenchClaw agent can spawn specialist subagents for complex tasks — a research agent, a code review agent, a data analysis agent — all orchestrated by the main agent.
The DenchClaw Stack Explained#
When you run npx denchclaw, here's what's actually installed:
OpenClaw (base framework)
└── dench profile
├── Gateway (port 19001) — AI agent runtime
├── Frontend (port 3100) — web UI
├── workspace.duckdb — your CRM data
└── Skills:
├── crm/ — CRM operations
├── app-builder/ — custom web apps
├── gstack/ — engineering workflow
└── browser/ — web automation
The dench profile is an OpenClaw configuration. It defines which model to use, which channels are connected, which tools are available, and what the agent's system prompt looks like.
Getting Started as a Founder#
Step 1: Install DenchClaw (and OpenClaw with it)#
npx denchclawThis handles all the OpenClaw setup for you. After this, openclaw commands are available in your terminal.
Step 2: Set Up Your CRM#
Tell the agent what you want to track:
"I'm a B2B SaaS founder. Set up my CRM with: investors (name, fund, stage focus, check size, status, last contact), pipeline contacts (name, company, role, status, source), and a deals table (company, value, stage, close date)."
The agent creates all three objects, generates YAML configs, and sets up the database views.
Step 3: Connect Your Channels#
For most founders, Telegram is the best primary channel:
- Message @BotFather on Telegram
/newbot→ name it → copy token- Tell DenchClaw: "Connect Telegram bot with token [token]"
Now you can message your CRM from your phone, during commutes, before and after meetings.
Step 4: Import Existing Data#
Where is your data now? A spreadsheet? Notion? LinkedIn? HubSpot?
"Import my contacts from ~/Downloads/linkedin-contacts.csv" "Import my investor CRM from Notion — I'm already logged in"
The agent handles the import, maps fields, and asks about anything ambiguous.
Step 5: Set Up Your Daily Briefing#
"Every weekday at 7:30am, send me a Telegram briefing: investors to follow up with, deals stalled for more than a week, and 3 top action items for the day."
Step 6: Learn What the Agent Can Do#
The best way to learn DenchClaw is to ask it things:
- "What can you do?"
- "How do I add action buttons to my deals table?"
- "Can you enrich my investor data with their latest portfolio companies?"
- "Build me a pipeline dashboard with charts"
The agent is your documentation. Ask it anything.
OpenClaw Features Founders Use Most#
Multi-Channel Messaging#
Your investors are on WhatsApp. Your co-founder is on Telegram. Your team is on Discord. One DenchClaw instance, connected to all of them:
# Connect WhatsApp
openclaw channel add whatsapp --profile dench
# Connect Discord
openclaw channel add discord --token YOUR_BOT_TOKEN --profile denchSubagent Workflows#
For complex, parallel tasks — like enriching 200 leads at once — the main agent spawns subagents:
"Enrich all my leads from last month — for each one, search their LinkedIn and Apollo profiles and update their role, company size, and LinkedIn URL."
The agent splits the work across parallel subagents and reports back when done.
The App Builder for Founder Tools#
Build custom tools for your specific workflow:
"Build me an investor CRM dashboard that shows: my target raise amount, amount committed, investors in diligence vs. soft circled, and a timeline to close."
The agent builds a .dench.app with live data from your DuckDB. Appears in your sidebar in minutes.
Background Research#
The browser agent can research things for you:
"Research all 12 investors in my diligence stage: their latest portfolio investments, fund size, check size range, and recent news. Update their records in my CRM."
The agent opens LinkedIn, Crunchbase, and their portfolio pages using your existing sessions and populates the data.
OpenClaw for More Than CRM#
Some founders use OpenClaw capabilities beyond CRM through DenchClaw:
Engineering workflow: The gstack skill turns DenchClaw into a structured engineering workflow — spec, review, implement, test, deploy — with specialist AI roles.
Customer success: Build a customer object, track health scores, automate check-in reminders, summarize support tickets.
Competitive intelligence: Build a competitors object, schedule weekly research runs, get briefings on competitor product updates.
Board meeting prep: Ask the agent to compile your board deck data — pipeline, metrics, team updates — from all the scattered places it lives.
Frequently Asked Questions#
Do I need to understand OpenClaw to use DenchClaw?#
No. DenchClaw handles all the OpenClaw complexity. But knowing it exists helps you understand the architecture and unlocks more advanced use cases.
Can I contribute to OpenClaw or DenchClaw?#
Yes. OpenClaw: github.com/openclaw/openclaw. DenchClaw: github.com/DenchHQ/denchclaw.
What does OpenClaw cost?#
OpenClaw is free and MIT licensed. DenchClaw is free and MIT licensed. You pay for AI model API calls (Anthropic, OpenAI, etc.) — typically $20-50/month for active use.
Is DenchClaw the only product built on OpenClaw?#
No. There's a growing ecosystem of tools built on OpenClaw. DenchClaw is the flagship, YC-backed product focused on the CRM/workspace use case.
Where do I learn more about OpenClaw?#
docs.openclaw.ai for the OpenClaw docs. discord.com/invite/clawd for the community.
Ready to try DenchClaw? Install in one command: npx denchclaw. Full setup guide →
