8:47 AM - Inbox triage
I pour my second coffee and open the Plumb dashboard. The little notification badge in my browser tab says 47 unread messages. Most of them are customer emails that came in overnight, which the AI agent flagged and sorted into three categories: questions, issue reports, and cancellation requests.
Sitting at my kitchen table in suburban Denver, I skim through the summarized queue. Today's signups dashboard shows 8 new contractor accounts created since yesterday afternoon. That's solid for a Tuesday. Revenue counter: three payments processed overnight, totaling $342. One of those is likely a free trial converting to paid. The other two are probably monthly renewals.
I pull open my Slack channel where the system sends alerts. There's a red flag notification from 6:23 AM: "Billing sync failed for contractor account - Stripe ID cust_J42xK. Manual reconciliation needed."
That one goes to my mental stack. I'll circle back.
First priority: the agent's email drafts. I click into Gmail's draft folder and scan what the system prepared. Three outreach emails, two customer replies, one cancellation response. I open the first outreach draft - it's to a contractor in Fort Collins named Marcus Webb who downloaded our YouTube tutorial on digital takeoffs last week.
The draft is solid. Personalized to his business (roof and gutter work), references the video he watched, asks about his current estimating process in his own language (not software jargon), and includes a soft demo link. I could send it as-is, but the tone feels a half-degree too formal. I edit the opening line from "I noticed you watched our recent tutorial" to "Saw you checked out our takeoff video last week - thought you might find this helpful." I hit approve and the system sends it.
The second draft is a reply to Carol Reyes at Reyes Family Plumbing. She asked if Plumb works on iPad, and her guys estimate jobs on the job site. The agent drafted a straightforward answer: yes, fully responsive mobile experience, syncs in real time, offline mode works. I add one sentence: "Most of our West Coast plumbers use it exactly that way - they said it cut their paper waste in half and made follow-ups way faster." A real customer benefit, not a feature list. Approve.
I skip the cancellation response for now and make a note to read the actual customer message first. That's where the real data is.
10:04 AM - The flagged conflict
My phone buzzes. Slack alert. The system flagged something unusual in the demo booking flow. I click the notification and see: someone submitted a demo request and, three minutes later, the system auto-created a Google Calendar invite with our sales time slot. But then the same person submitted another demo request in a different time zone. The system generated a second conflict.
This is the kind of edge case where automation breaks if you're not paying attention. I click through to the customer record: David Park, small commercial HVAC contractor in Boulder. He filled out the first form at 9:41 AM, then again at 9:44 AM. Probably didn't see the confirmation, thought the form didn't go through, and resubmitted.
I pull up his email and send a quick message: "Got your demo request, David. I scheduled you for Tuesday at 2 PM Mountain Time - that shows up on your calendar already. You're all set. Looking forward to showing you how other HVAC crews are cutting estimation time by 60%." I delete the duplicate calendar invite manually in Google Calendar.
That's the friction point nobody talks about when they sell you "AI runs your business." The AI catches 95% of what needs to happen, but that last 5% breaks the customer experience if you're not on top of it. I toggle back to my notes and add a task: "Improve demo form logic to catch duplicate submissions within five-minute window."
11:47 AM - A conversion closes
I'm halfway through reviewing this week's pipeline when an email lands in my inbox from someone I demoed with last Thursday. James Chen owns a mid-size electrician outfit in Aurora, employs about 12 people. During the demo, he asked smart questions about API integrations with his existing QuickBooks setup. Today he's writing: "Alright, we want to try it. How do we get started. What's the onboarding like."
I don't send him to a bot or a generic onboarding sequence. This is a customer who's going to likely stick around (he asked integration questions, which means he's planning to use it). I write back: "Great, James. I'm going to walk you through this personally today. Can you do 15 minutes this afternoon. I'll set up your billing in Stripe right now and send you a Zoom link. One thing up front: I'll import your last month of projects into the system so your crew can see the historical pattern. Saves us time getting you productive." I hit send and make a note to action his account setup before 1 PM.
12:31 PM - Metrics check and the churn conversation
Lunch is at my desk. Grilled cheese and tomato soup. Slack on one side, Plumb dashboard on the other.
This week's numbers so far:
New signups: 34
Demos booked: 18
Conversions to paid: 6
Active paying accounts: 247
MRR: $29,640
Today's pipeline value: $18,500 (15 opportunities in various stages)
That $18,500 is mostly warm leads - people who've had demos or initiated the trial. If we close half, we're looking at a $9K week, which would be on pace for the year.
One metric I watch closely: churn. I notice in the dashboard that someone canceled this morning. I pull the details. It's a smaller operation, Basic plan at $29/month, been with us for eight months. The system flagged the reason they gave on the way out: "Too much manual entry still. We need something faster."
This one stings a little because we actually do have faster workflows, but they probably didn't find them or didn't have someone to show them the right way. I fire off a quick email to her (Donna Martinez at Martinez Roofing) with two specific workflows that might have helped: bulk job import and the batch estimation tool. I offer a brief call if she wants to reconsider. I don't expect her to come back - the tone of her cancellation message was pretty final - but I'd rather go down swinging than assume we were the wrong fit.
2:14 PM - Demo with James and a bug
James and I jump on the Zoom call at 2 PM. His QuickBooks database is bigger than average - hundreds of past projects. I show him how the integration maps estimate line items into his existing QB structure, so his accountant doesn't have to re-enter anything. His face changes when he sees it. He says, "That right there saves us two hours a week." We talk through his team's mobile workflow, and I walk him through the iPad interface. He's sold. I process his credit card on the Stripe dashboard while we're on the call (he gives me his details verbally over Zoom, and I enter it in our payment form). First month processed. I send him a welcome email with documentation links and let him know I'll check in Friday to make sure the QB sync is working.
After the call, I try to test something on his account in the admin interface and notice a bug: when you create an account for an existing contractor business (one that's already in the system via a previous user), the system doesn't always merge the history correctly. It's not a critical issue - it doesn't break anything, just clutters the data view. I open a Linear ticket, mark it as medium priority, describe the steps to reproduce, and assign it to myself for next week. This is not something I can fix in five minutes, and it's not urgent, so I document it and move on.
4:42 PM - Stripe reconciliation and pipeline review
I circle back to that morning alert. The billing sync that failed for that contractor. I dig into the Stripe dashboard, pull up customer cust_J42xK, and see that they made a payment, but our system didn't record it. Likely a webhook timeout or a network blip. I manually update the account status in our database to "paid" and send the contractor an email letting them know their service is active. No explanation needed - they're probably already using it. But I flag this pattern: we're seeing about one sync failure per week now. Something to dig into with more urgency as we scale.
I open a second browser tab and pull up our week-to-date pipeline in the CRM view. Eighteen demos scheduled for this week, six already completed. The conversion funnel from demo to paid is holding at about 35%, which is solid for a SMB SaaS product in the trades. Most of the outreach happens through the automated system - initial emails to people who've interacted with our content - but every single one of the closed deals involved at least one manual conversation.
I see the names: James Chen (just closed today), Melissa Ortiz (electrician in Pueblo, scheduled tomorrow at 10 AM), a group demo with a small GC firm that has seven crew members. That one's valuable - if it closes, it's not just $120 a month, it's $120 times seven.
I make a note: tomorrow I need to prep a custom walkthrough for the GC group. They're going to want to see team workflows and job delegation. That's not in our standard demo script.
6:08 PM - Closing down
I close Slack, export today's activity log, and jot down three quick notes in my running reflection doc:
The demo booking duplicate submission happened because the form didn't give visual feedback. Needs a UI fix.
Stripe sync failure happening once a week suggests we're hitting some rate-limit or timeout. Get the engineer to look at webhook retry logic.
Donna Martinez's churn stung. We had the features she needed. Next time someone mentions "too slow," I need to proactively show them the faster workflows in the first demo.
My laptop closes at 6:12 PM. The day brought in three new paying customers (James, and two from the auto-renewal cohort), eight new signups, $342 in revenue, and one cancellation. The pipeline is healthy. The operation is moving.
It's real work. I'm not sitting back watching an AI make money while I sleep. I'm making decisions, fixing customer friction, reviewing AI-generated content for tone and accuracy, catching the five percent of things that fall through automation. But the automation is the thing that makes this workable. Without the agent handling email sorting, demo scheduling, and outreach, I'd be drowning in 200 emails a day. Instead, I'm seeing 47, and I'm making faster decisions because the system presented them clearly.
By Friday, if the GC group converts, I'll have added probably $500 in new MRR this week. At the current burn rate, that stacks up. This is a real business. It's work, but it's mine.