Automation
Chatbots and CRM automation that feed the loop.
The gap
A chat widget on the site that nobody tracks. Leads scored by whoever picks up the phone. A CRM that is updated on Fridays, if at all, so nothing useful ever goes back to Google.
What we build
Lead scoring before upload. Every enquiry passes through an LLM with one question and a rubric: is this a real lead, and how good? The score sets the conversion value, decides whether it goes back to Google at all, and routes the lead to the right person. This is what took a lead gen client from about €50 to €20-30 per lead. The ads got better because Google was told which leads were worth having.
Chatbots that capture rather than deflect. Built on a fast, cheap model tied to the CRM (Claude Haiku with HubSpot on a recent build). The bot qualifies, books or hands off, and writes the conversation into the CRM record. The required EU AI Act disclosure is built into the first message.
CRM workflows in n8n. Stage changes trigger uploads. New customers join the exclusion audience. Quotes that go quiet get a follow-up. The CRM stays current because the automation does the boring part.
All of it self-hosted n8n, so the workflows are yours and there is no per-task bill.
What you get
The scoring workflow with the rubric written in plain English so you can change it. The chatbot, its disclosure copy and a transcript log. The CRM workflows, documented. Monitoring that tells us when something stops running.
Related: CRM feedback loop · Tracking
FAQ
Will an LLM misjudge leads?
Sometimes. It is checked against what the CRM later says happened and the rubric is adjusted. It is already better than scoring by whoever answers the phone.
Do we need HubSpot?
No. n8n connects to most CRMs. If yours has an API or even an email inbox, it works.
What about data protection?
Lead data stays in your CRM and your n8n instance. The LLM sees what it needs to score and nothing is retained by it.