8:42 AM - Inbox Triage
I open my laptop on the kitchen table with a cold coffee from an hour ago. The Engagement AI dashboard loads. There's a Slack notification waiting from the system: "4 customer emails sent overnight, 1 error flagged, 12 new signups." I click into the inbox. Tuesday mornings are usually like this. The AI has been drafting re-engagement emails all weekend and last night, and my job on Tuesday is to review them, fix the ones that miss the mark, and hit send.
First email is to Marcus Zhang at Zhang's Juice Bar. His store had two orders in January, nothing since. The AI drafted an email offering a 15% discount on his next smoothie order plus a free add-on if he buys within 7 days. It's good. The tone is right, the discount isn't too aggressive, and it's personalized with actual product names from his store. I approve it.
Second one goes to Priya Sharma at Sharma Athletics. Her retention rate has been sinking for three weeks. The AI flagged her as a churn-risk in the segment report last Friday. It drafted an email asking what went wrong, offering a free month of her subscription to come back. This one is too apologetic. I delete it and write my own version in Gmail: a simple note saying we noticed she'd gone quiet, asking what she'd use in a re-engagement offer. It takes me four minutes. I hit send.
Third draft is different. This one isn't a re-engagement email. It's a thank-you note to Carol Reyes at Reyes Family Practice. Carol signed up on March 18th for a four-month trial. The AI noticed that she's now booked 47 new patient follow-ups through email campaigns that the engine ran. The thank-you email is warm and specific. I read it twice. Then I approve it without changes and add a personal note on top: "Carol, I'm one of two people running this company. Your practice helped me see that Engagement AI works best when the business owner actually talks to customers. Thank you." I send it.
By 8:57 AM, I've sent five emails.
10:15 AM - A Flagged Conflict
I look at the Linear board where the AI logs edge cases it can't handle on its own. There's a new ticket from yesterday evening. A customer, Jake at Jake's Vinyl Records, has two different email addresses registered in our system. One has been inactive for eight months. The other has been active for two weeks. The AI is asking which one to target with next week's campaign. This is the kind of thing that happens when someone signs up, forgets their password, creates a new account instead of resetting it.
I go to his profile in the Engagement AI dashboard. I can see the history. Both accounts have real activity. I don't want to email Jake twice, and I don't want to alienate him by choosing the wrong email. I open Gmail and send him a short note: "Hey Jake, I see you have two accounts with us. Which email do you want to use going forward?" I flag the Linear ticket as "awaiting response."
This is where the product hands off to me. The AI is great at writing marketing emails and spotting patterns in churn. But it can't navigate the gray stuff. I'm not frustrated by that. It's why the product costs $100 a month, not $1000 a month. The owner still has to be the owner.
12:30 PM - Lunch and the Metrics Check
I make a sandwich and pull up the weekly metrics in a spreadsheet I sync from Stripe. It's Tuesday, so this is my standing check.
Week to date (Monday to now):
- 28 new signups
- 7 lost customers (churn)
- $3,240 MRR from existing customers
- $620 in new trial revenue (new signups at $30 for two weeks)
- 3 demo meetings booked (I can see these in my Slack #sales channel, where the sales bot posts lead notifications from LinkedIn)
The churn number is the one I linger on. That's up from an average of 4.5 per week. One of those is Priya Sharma, the Athletics subscription service. I made a mental note this morning that her re-engagement email wasn't working. She probably needs a personal call, not an automation. I make a note to reach out Thursday.
The demo number is good. Three prospects in a single week is on track for 12 this month. At a 25% close rate, that's three new customers by end of May. Nothing is guaranteed, but the cold outbound to LinkedIn has been working since I stopped writing the messages myself and started using the product to draft them.
I eat the sandwich and close the spreadsheet.
2:08 PM - A Bug and Two Emails
I check Slack. There's a message in #support from a customer named Wei Chen at Wei's Ceramics Studio. She says her customers are getting re-engagement emails about products she no longer sells. The product categories in her store changed three weeks ago, but Engagement AI is still pulling from the old data.
This is a bug, or at least an edge case in the integration. The system is supposed to sync product data from Shopify every 24 hours. I open the admin dashboard and look at Wei's sync logs. The last sync was 28 hours ago, and it's failing silently. The Shopify API token probably expired.
I ping our developer in Linear with the details and mark it as "urgent." Then I write an email to Wei: "We caught a sync error on your end. I've got our team on it and will update you by EOD Wednesday. In the meantime, I've paused your re-engagement campaigns so you don't send any more messages about discontinued products." She replies within two minutes with a thumbs-up emoji. That feels good.
The second email comes in from a prospect, Tom at Tom's Coffee Roasters. He's asking if Engagement AI can integrate with his SMS platform. We don't support that yet. I write back honestly: "Not yet. We're SMS-native on the roadmap for Q3, but I can't promise that timeline. Are you interested in email for now, or would you rather wait." He doesn't respond, which probably means he'll wait. That's fine. I'd rather lose a deal to honesty than win it with a promise I can't keep.
4:30 PM - Pipeline Review
I open Pipedrive and look at the three prospects from this week. One of them, Noor at Noor's Skincare, had a demo on Monday. She liked the product but asked about pricing flexibility. I haven't followed up yet. I spend 20 minutes putting together a pricing note specific to her use case: volume-based discounts if she hits 500 active customers. I send it over.
The second prospect, Tom from the SMS question earlier, is probably dead for now.
The third, Jamal at Jamal's Fitness, had his demo last Friday. He said he'd decide by Wednesday. It's Tuesday. I send him a short note: "Wanted to check in. Do you have any other questions before you decide." Again, real work. Simple work. But necessary work.
I think about the metrics from lunch. If two of these three close, that's $200 to $300 in new MRR by the end of the month. At the rate we're operating, that compounds. It's the thing that keeps me going some days, the math of it, even if it's slow.
6:15 PM - Wrap
I close the dashboard. I've been in the chair since 8:30 AM with one break for lunch. I wrote maybe 15 emails today, reviewed 8 emails that the AI drafted, fixed one sync bug (well, reported it), handled one edge case, looked at metrics, and followed up on three deals. None of it is complicated work. All of it matters.
The thing that strikes me as I shut down the laptop is that I'm not running Engagement AI by standing back and watching it work. I'm running it by being in it, by catching the weird cases, by sending the personal emails that the templated stuff can't handle, by watching the numbers and asking why. The AI is an amplifier. It wrote probably 40 emails this week that I didn't have to write. But I still wrote 15 myself. I still review everything. I still own the customer relationship.
Churn is up. The Shopify sync broke. Two of three demo prospects might not close. These are real problems with real consequences. But the business is moving. Four months in, and I'm not looking for a way to automate myself out of this job. I'm looking for a way to do more of it, faster, without losing the parts that are working.
I'll be back here tomorrow morning.