SaaS

How Kumar Abhirup Built DenchClaw, a Local AI CRM on OpenClaw

Built by Kumar (Dench.com) · · Source: producthunt

Quick answer

Kumar Abhirup and Mark Rachapoom of Dench.com built DenchClaw, a free open source CRM that runs on your Mac on top of OpenClaw. You edit the CRM in plain language, and agents enrich leads and do outreach through your browser.

Result: According to the Product Hunt page: 178 upvotes and #18 for the day. Kumar says the earlier Ironclaw version had about 1.3k GitHub stars and that the team uses it daily.

What they built

Kumar Abhirup and Mark Rachapoom, co-founders of Dench.com, built DenchClaw, a free open source CRM that runs locally on your Mac on top of OpenClaw. The Product Hunt listing describes it as a managed OpenClaw framework for knowledge work, with CRM automation and outreach agents.

According to Kumar's comments on the launch, Dench.com went through Y Combinator in Summer 2024 and used to be an agentic workflow company. The project was first called Ironclaw and was renamed after confusion with NearAI's Ironclaw. Kumar compares OpenClaw to early React, a basic building block, and says DenchClaw is an opinionated framework on top of it.

How it works

Kumar gives a fair amount of setup detail in the comments:

  • Install: one command, npx denchclaw. It creates a separate OpenClaw profile called dench, runs a gateway on the 19001 port range and serves the frontend at localhost:3100. He says you can add it to the Dock from Safari as a PWA.
  • Data: the CRM is stored in DuckDB and the local file system. Views, filters, column toggles and calendar or Gantt views are saved as files, so OpenClaw can edit them when you ask in plain language.
  • Lead enrichment: subagents enrich leads and stream their progress back.
  • Browser work: the agent runs in a visible browser session using your own Chrome profile or a new one. Kumar says data from Notion and HubSpot can be imported through the browser.
  • Syncing: iCloud for one person, GitHub for a team, or a self-hosted VM for a whole company.

When a commenter asked for a first task to try, Kumar suggested finding LinkedIn contacts by title, category and city, adding them to the CRM and drafting outreach messages.

The result

The Product Hunt page shows 178 upvotes and a rank of #18 for the day, with no written reviews yet. Kumar says the earlier Ironclaw version had about 1.3k GitHub stars. He says he uses DenchClaw every day, including as a coding agent, and Mark says the whole team uses it daily.

How to build something similar

  • Keep your CRM data somewhere an agent can read and write directly, such as a local database plus plain files for views and filters.
  • Give the agent its own profile or workspace so its CRM work stays separate from your other agent setups.
  • Start with one narrow task, like building a small lead list and drafting messages, and review the drafts yourself before anything gets sent.
  • Run browser automation in a window you can watch, so you can see what the agent is doing and stop it if it goes wrong.

The original post

Kumar Abhirup · View on producthunt

Want to build something like this?

Join the Shipping Skool community free. Paid members get live bootcamp calls every week.

Join Shipping Skool — Free

More use cases

← All AI agent use cases