DenchClaw
DenchClaw is a locally-run AI CRM that combines a conversational database interface, browser automation, and pipeline management in a single tool.
About
DenchClaw is a locally-run AI CRM that combines a conversational database interface, browser automation, and pipeline management in a single tool. It runs on localhost:3100 via npx denchclaw and uses natural language to translate queries into SQL against a local DuckDB database. The sales focus is practical: it automates LinkedIn prospecting using your existing Chrome session, handles lead enrichment, and manages outreach sequences through a kanban-style pipeline. Unlike cloud CRMs, all your data stays on your machine.
Technical founders and solo sales operators who want CRM-level automation without paying for another SaaS subscription and are comfortable with a localhost setup. Works especially well for outbound-heavy workflows targeting startup ecosystems or niche verticals where LinkedIn prospecting is central. <!-- Screenshot pending -->
Pros & Cons
Pros
- check Runs entirely on localhost — no SaaS subscription, no data leaving your machine, no vendor access to your contacts
- check Natural language to SQL interface means non-technical sales operators can query and segment leads without writing code
- check Uses your existing Chrome authentication, so LinkedIn automation and web scraping work without separate login flows
- check Cron-based scheduling handles recurring tasks like weekly reports and automated follow-up sequences without manual triggers
- check Access to 58,000+ skills via the integrations marketplace extends functionality well beyond the base CRM
Cons
- close Requires Node 22+ and comfort running tools via the terminal — not viable for non-technical users
- close Running locally means no mobile access, no shared team view, and no built-in backup unless you configure one
- close LinkedIn automation carries account risk, as LinkedIn actively detects and restricts bot behavior regardless of the tool used
- close Pricing and support model are not disclosed, making it hard to evaluate for longer-term use