8:42 AM - Inbox triage
I open my laptop at my kitchen table. Coffee in hand, cold from sitting there for twenty minutes while I checked my kids' backpacks. The Slack notification badge shows seven new messages. Investor AI's #alerts channel first: three flagged drafts overnight, two successful generations, one billing alert. Normal Tuesday. One of the flagged ones is marked high priority.
I open the Investor AI dashboard. Week-to-date: 7 new demo signups, 3 of which became qualified leads in the pipeline. One of those is actually looking solid, asking detailed questions about compliance. I make a note to follow up with them by Thursday at the latest. Today's forecast shows one expected monthly renewal hitting my Stripe dashboard later, probably around 3 PM when the batch runs.
The Slack alerts are mostly routine. One is different: a customer suitability letter that the system flagged as potentially non-compliant. It's using "aggressive growth strategy" language without backing recommendations against the stated risk profile. Good catch by the algorithm. I flag it to review this morning.
My email shows 23 new messages. I filter to VIPs only. Two are actual customers. One is Carol Reyes at Reyes Family Practice - a two-person RIA that came aboard three weeks ago. She wrote: "The draft portfolio review summary you sent yesterday saved me an hour. This is exactly what I needed." That feeling is why I'm doing this. I screenshot it and post it to Slack.
The other is from Marcus Chen at Axiom Wealth. He's asking a billing question: whether the platform can handle multiple sub-advisors on one account, or if each one needs their own seat. It's a product limitation. I know the answer and I know it's not ideal. I'll need to respond to this manually, probably suggesting a workaround for now. This is the kind of thing that eats forty minutes of my day that the AI can't solve for me.
10:15 AM - A flagged conflict
I navigate to the flagged suitability letter. It's from James Liu at Beacon Advisors, a new customer who signed up five days ago. The AI drafted a recommendation for a dividend-focused portfolio rebalancing. But here's the problem: the client profile shows moderate risk tolerance and a time horizon of seven years. The system correctly flagged that the draft doesn't explicitly tie the recommendation back to those constraints.
I read the AI's draft twice. It's not wrong, exactly. But it's sloppy. A compliance officer would ask, "Where's the documentation showing this aligns with the client's stated risk tolerance?" In financial advice, sloppiness on this specific question is expensive.
I don't reject the draft. Instead, I mark it for revision and leave a comment in the platform: "Rewrite sections 2 and 3. Explicitly state how dividend income aligns with moderate risk profile and seven-year horizon. Reference client's stated goals from intake form." I send this back to James with a note asking him to review the revised version when it's ready.
This is the workflow that works: the AI drafts. I catch the edge cases and the compliance gaps that are specific to each customer's situation. I'm not reviewing every draft, but I'm reviewing the risky ones. The system learns which categories get flagged most often.
11:35 AM - Customer email and approvals
Marcus's billing question is sitting in my inbox. I need to actually answer it. The short answer is no, we don't support sub-advisor seats yet. The medium answer is that he can work around it by creating separate client groups and assigning them in his workflow. The real answer is that this is a feature request I'm going to hear from more people, and I'm making a note in Linear to prioritize it in the next sprint.
I write back to Marcus: "We don't have sub-advisor licensing yet, but I can walk you through a workflow that gets you close. Can you grab 15 minutes this week? I want to understand how you're structuring teams so I can make sure we build this the right way." I schedule him with a Calendly link. This is operator work. This is also how you keep customers.
I shift back to approvals. Three suitability letter drafts are sitting in my queue, all pre-flagged as compliant. I read each one. Two are solid. One is from a customer I trust, Sarah Mitchell at Silver Birch Wealth Management. She's been with me for four weeks and has sent clean feedback twice. I approve the three drafts, and Investor AI sends them to the customers' email addresses for final review.
12:50 PM - Lunch and the numbers
I make a sandwich and pull open my Stripe dashboard. This is the real-time revenue view. I'm looking at recurring subscription revenue, not one-time signups. Four active subscriptions showing on the dashboard:
- Reyes Family Practice: $199/month
- Beacon Advisors: $179/month
- Silver Birch Wealth Management: $249/month
- A fourth customer, Cornerstone Financial Planning, that came on last week: $169/month
That's $796/month recurring right now. Today's renewal should push it to $995. The math I did when I decided to build this: 180K ARR means roughly 60K first-year customers, or 15K per quarter. But I'm targeting the smaller end of the market. A hundred paying customers at $150/month is 180K. A hundred feels achievable in year one if the motion works.
I pull up my pipeline. Seven qualified prospects. Three are warm (they responded to outreach, asked questions). One is very warm (shared a specific use case). Three are cold (they opened my email, nothing more). I've done eighty outbound emails this week. The conversion math is brutal, but it's also predictable.
I check the Google Sheet where I'm tracking retention. Three customers, four weeks. No churn yet. But I'm watching.
2:30 PM - The edge case
Slack pings. It's Carol from Reyes Family Practice. She's asking if the system can handle a situation where a client's risk profile changed mid-quarter, and she needs to update recommendations retroactively without re-running the entire meeting summary.
This is not a bug. This is a real operational question that doesn't have a clean answer in the product. I could tell her the system doesn't handle it. I could walk her through a manual process. Or I could ask her what she needs, understand the pattern, and prioritize fixing it.
I call her. Thirty seconds in, I realize this is actually a compliance documentation issue more than a product issue. She needs a clean audit trail showing when the recommendation changed and why. That's not something the AI draft captures automatically.
We talk for twelve minutes. She tells me this is probably happening quarterly for her. That means it's not edge case. That means it's a feature. I add it to Linear with her name on it and commit to building it within two weeks. She seems satisfied. I feel good about this conversation.
4:15 PM - Pipeline and manual labor
I spend an hour on outbound. Cold email doesn't scale if I'm writing personal emails, but these aren't cold anymore. They're warm-ish. I've got seven prospects to re-engage. The ones that opened my first email and didn't respond. I write brief, specific follow-ups.
One of them is actually from a referral. A customer at an RIA I called last month mentioned my name to Marcus at Axiom - the one with the billing question. That's good validation. I note that in the CRM.
My Stripe dashboard refreshes. The expected renewal came through. Four-fifty-nine seconds later, my Slack alerts channel shows it: $249 payment captured. That's from Silver Birch's monthly renewal. It's a small number, but it's predictable revenue. It's validation that people see value enough to keep paying.
5:45 PM - A small fix and one manual draft
There's a bug I found last week that's been bothering me: when a customer uploads a new client file, the system sometimes duplicates the compliance checklist. I spend thirty minutes in the codebase, find the issue (a missed condition in the merge logic), and push a fix. I test it locally. Works. I deploy it to production. No crash.
James Liu from Beacon Advisors replies to my flagged suitability letter. The AI revised it. I read the new version. It's better. I approve it and send it to him with a note: "This version documents the alignment explicitly. Should clear any compliance review now." He responds within three minutes: "Perfect. This is exactly what I needed."
That exchange took eight minutes of my time. It's the kind of thing that scales poorly but matters enormously to the customer.
I get one more email. It's from a prospect I emailed four days ago. They want a demo. I check the calendar, find a slot, and send it back. Now I have a demo scheduled for Thursday at 10 AM. That's one qualified conversation I wouldn't have had if I wasn't in the email thread myself.
6:15 PM - Closing the laptop
I close the Investor AI dashboard. Check Slack one more time. Alerts are quiet. Revenue for the day is confirmed. The four refunds I was dreading from churn didn't show up, which means churn isn't accelerating. I can live with one placeholder renewal tomorrow.
I've been doing this for five weeks now. The narrative I tell myself is different from the narrative I would've written a month ago. I thought the hard part would be building the product. The hard part is the operator work. The AI drafts, and drafts well. But the compliance calls, the customer follow-ups, the bug fixes, the pipeline scrubbing, the strategic decisions about features - those are still me. The AI is the amplifier. I'm the filter.
The thing that worked today: Carol's call. Understanding a real operational pattern from a customer and committing to fix it. The thing that felt hard: Marcus's billing question and realizing we're not ready for some real use cases yet.
I tell myself to remember this on the days when the metrics look small. Ninety-six dollars in recurring revenue today. Four active customers. Seven in the pipeline. It's real. It's movement. And unlike a job, I don't have to do what someone else tells me. I'm building something.
I close the laptop at 6:17 PM. Tomorrow is Wednesday.