I built an AI assistant that runs my office
A few weeks ago I shared how I built a one-step enrollment system for my lawn care program — client enters their info, credit card goes on file, and they're enrolled. No quote approval, no back-and-forth. Automated sequences handle the welcome email, welcome text, and contact setup. One step, done. Coming from a background as an electrician, I tend to look at everything as a circuit. If there’s a break in communication or a 'loose wire' in my lead flow, the whole system fails. I decided to stop fighting the mess and started 'wiring' my office the same way I would a complex panel. I built an AI operations system that runs alongside Jobber. Every text, phone call, email, voicemail, missed call, and website form submission automatically gets logged to a centralized database through a series of Zaps. Every client has one record, one timeline, and one clear next action at all times. Every morning before I head out, I run a 5-minute briefing with an AI assistant. It reads the full client database through an MCP server — which basically means the AI has live access to every client interaction in real time. It tells me who contacted me overnight, who's waiting on a response, who's going cold, and what I should do next for each person. It drafts the messages. I review, edit if needed, and send. I also set up an AI receptionist on my business phone line. It answers calls, can answer common questions about services and pricing, takes client information, and transfers calls when needed. It can also send texts to the caller during or after the call — like a direct link to the enrollment page or the resources section on my website. It handles multiple calls at the same time. No more missed calls going to voicemail. The tools: Jobber for jobs, scheduling, and service history. An MCP-connected database for the client timeline. Zapier to connect everything. An AI assistant for daily briefings and client communication. An AI receptionist for inbound calls. Jobber stays at the center — it's my source of truth for every job, every visit, every quote. The AI layer sits on top and makes sure nothing falls through the cracks between Jobber and everything else. I'm planning my first hire this season. Not because I'm behind — because the systems are handling the admin load well enough that I can focus on growing. The AI doesn't replace a person. It replaced the office work I used to do at 10 PM after a full day in the field. If anyone's curious about how any of this works, happy to answer questions. I'm not selling anything — just sharing what I've built because this community helped me think through a lot of it.Anyone else built a direct MCP connection between Claude AI and Jobber?
I set up a direct MCP integration between Jobber and Claude AI — meaning Claude can query my Jobber account in real time and pull live data like client counts, job status, and invoices on demand. Took some work to get running but it's live. Current limitation is it's read-only. Jobber's MCP tools are still limited on the write side, which is where it would get really powerful — imagine AI-assisted job creation, status updates, or dispatching without touching the app. Curious if anyone else has built this connection, and more importantly, if anyone has figured out how to get write access working.
