Day 1 operating Funeral AI · Wishdeal Factory

8:42 AM - Inbox triage

I open the dashboard at 8:42 on Tuesday morning. My coffee is still hot. The funeral industry doesn't sleep much, and neither do I.

Overnight, the system logged four new arrangements: two intake forms submitted around 6 PM yesterday, two more that came through at 3 AM from funeral homes on the West Coast. The AI agent had already drafted initial intake confirmations for all four - basic acknowledgment emails with a next-step outline. I scan the drafts in the Gmail thread view. Three of them are solid. The fourth one, from Sandra Martinez at Martinez & Sons Funeral Home in Albuquerque, has a tone problem. The agent offered condolences in the standard language, but Sandra's intake form flagged this as a suicide loss. The draft email felt generic for that context. I manually rewrite it, softening the language, removing the perfunctory "we understand your loss" and replacing it with something more specific: "We know this was unexpected and painful. We're here to handle the administrative details so you can focus on your family." Hit send.

I glance at Slack. Three alerts came overnight. One is routine - a confirmation that the nightly backup completed. Two others are flagged arrangements: customers with more than one open case. That shouldn't happen. I make a note to investigate those after I move through the queue.

10:15 AM - The conflict

By 10:15, I'm into the week's backlog. There's a woman named Carol Reyes - owns Reyes Family Practice in Phoenix - who's been with Funeral AI for six weeks. She's a good customer. Reliable, responsive. This morning, the system flagged a conflict in her arrangement: a family submitted a service date (Saturday at 2 PM) but Carol's notes show a direct cremation, no service. The AI agent had caught the discrepancy and flagged it in the dashboard rather than sending an automated response. Good. This is exactly what I need to see.

I click through to Carol's intake and see the issue immediately. The family submitted an online form yesterday saying they wanted a memorial service Saturday afternoon. But Reyes's notes from the initial phone call, recorded last week, say "direct cremation to ashes, family scattering at later date." The family's form also indicated they checked "cremation" in the initial section but then checked "service" in the follow-up section - probably user error.

I send Carol a Slack message instead of email. She's responsive on Slack. "Hi Carol, saw a potential mismatch on the Martinez family case. Their form says service Sat at 2pm but your notes have them direct cremation. Can you clarify what they actually want. Happy to draft the right follow-up email once we confirm." She replies in four minutes: the family definitely wants direct cremation, no service. They must have clicked the wrong button. I then draft the clarification email to the family, have the AI agent refine it, review it, and send it out. By 10:47, it's resolved and confirmed. This is the work that makes the difference. The AI finds problems. I solve them.

12:15 PM - Metrics check

I break for lunch but keep one eye on the dashboard. I open my Stripe account in a separate tab while eating a sandwich. Today's revenue: $847. That's solid for a Tuesday. I can see thirteen of my customers paid their invoices last night, three are pending due by end of week, and one shows as 30 days past due. I click into that account. It's Miller Funeral Home in Des Moines. I remember them - they've been with me for eight weeks but the owner said something about switching their workflow back to their old system. This is likely a churn. I make a note to call them this week. The churn rate is still low enough that each customer matters.

I pull the dashboard metrics view. Week to date: 31 new arrangements processed, 18 fully completed and billed out, 13 still in flight. New customer signups this month: 6. That's tracking toward my 5-8 target. Pipeline - the prospective customers I've demoed to - shows 14 conversations in progress, 4 that are late-stage within two weeks of a yes/no decision. One of those is the director at a 200-family funeral home in Colorado Springs. That close would add about $1,200 a month to recurring revenue. It matters.

2:08 PM - The edge case

At 2:08, I get an email from James Chen at Chen Family Funeral Services in Portland. He's been a customer for two weeks and is still learning the system. His question: one of his families wants to add two additional services beyond the "standard" arrangement package the AI had outlined. The family wants to add a graveside service and a separate memorial dinner arrangement.

This is the kind of thing my system isn't built to handle natively. Funeral arrangements are deeply custom, and the AI agent can help standardize the common patterns, but outliers exist. I need to think about this one.

I reply to James. I explain that the system handles the core arrangement workflow, but special requests like a multi-event package need some manual configuration. I walk through what we'd do: break the arrangement into three separate line items in his system, each with its own timeline and cost. I draft a simple one-page worksheet in Google Docs, shared with him, that shows how to structure it. I also post our Slack support channel and ask if any other customers have done multi-event arrangements. No one has. So I do something I should have done earlier. I add a new prompt to the AI agent: "If an arrangement includes more than two distinct events, flag it for manual operator review." I push that change live in the admin settings. Now the next time someone submits something complex, I'll see it proactively instead of waiting for a customer email.

4:30 PM - Pipeline and positioning

By 4:30, I'm thinking about next month. I've got four conversations in flight that feel warm. One of them is Derek Thornton, who runs a three-location funeral home chain in Ohio. I met him at the NFDA member event last month - he was sponsoring the state association booth and seemed genuinely interested in streamlining operations. I've sent him two demos and a case study from a similar-sized operation in Michigan. He hasn't said yes, but he hasn't said no.

I decide to send him a personal note instead of a templated follow-up. I open Gmail and write something genuine: "Derek, following up on your demo from last week. I know you're evaluating a few options. What I'd offer is that we can start with just one of your three locations - prove the value there, then scale to the other two if it makes sense. No commitment to all three at the start." I send it. It's a small thing, but it's the difference between feeling like a vendor and feeling like a partner trying to solve a real problem.

5:45 PM - A good note and the day wind-down

At 5:45, I see an email from a customer I'd almost forgotten about. Theresa Williams at Williams Funeral Home in Nashville. She's been with me for twelve weeks, one of my first customers. She wrote: "Just want to say thank you. Your system saved me about five hours a week on paperwork. I'm hiring an additional staff member because I have time to focus on families instead of forms. It's making a real difference."

I read that twice. That's why I'm doing this. That's the legitimate output of the work. Not every day brings a note like that, but every day brings a little bit of the underlying truth: I'm helping people spend less time on paperwork and more time on what actually matters.

I close my laptop at 6:13. Today felt good. I handled a tone-sensitive email myself. I caught a conflict and resolved it quickly. I fixed a process gap by updating the AI agent's instructions. I clarified a tricky edge case for a customer. I moved one prospect closer to close with a personal note. I lost one customer to churn, but I'll call Miller tomorrow and try to understand why.

The business is real work. The AI is an amplifier, not a replacement for judgment. I'm not sitting back watching a robot earn me money. I'm actively managing relationships, solving problems the system can't solve alone, and building trust with funeral directors who are busy enough without worrying about their technology. This month, if we hit 7 new customers, we'll be at 38 customers, about 13,500 in monthly recurring revenue. The numbers are moving in the right direction. And I'm still running this alone, mostly in the morning and early afternoon before I hand some of the operational details to my part-time support person next month.