8:42 AM - Inbox Triage
I open my email and see the Counsel digest hit at 6:47 a.m., right on schedule. Three overnight intakes. My coffee is still warm. I scan the summary line for each: commercial dispute in Oregon, personal injury in Arizona, matrimonial in Massachusetts. The agent has already formatted them into the intake form template. I pull up the admin dashboard to see the full detail.
The Oregon call came in at 11:23 p.m. from Marcus Webb, potentially a commercial dispute involving a contractor and a small manufacturing firm. The call transcript is there. I skim it. The agent did well on the jurisdiction question, caught that Webb mentioned "opposing counsel in Portland," and asked the right follow-up about whether the firm was already retained. Clean intake. I check the conflict flag: green. No matches in the adverse-party database I maintain. I open Stripe to log the event - the system credited $25 to this firm's account for the successful intake capture. One less hour of ringing phones for Webb's paralegal to answer at night.
The Arizona case is trickier. Sandra Ochoa, a solo practitioner, called about a slip-and-fall. The agent captured it well, but flagged it yellow. Why. I click through. The opposing business name is "Phoenix Wellness Center." I have a note in my database that Ochoa represented Phoenix Wellness Center three years ago in an unrelated employment matter. It might be nothing. It might be a conflict. The agent did the right thing by flagging it.
I open a Gmail draft to Sandra. "Hi Sandra - the overnight intake system flagged a potential prior relationship with the defendant. You represented Phoenix Wellness Center in 2021 on employment matters. You know this business better than I do. Should we take the call, or flag it to decline? Let me know, and I'll update the retainer status." I send it. She'll see it before 9 a.m.
The Massachusetts case is worse.
9:15 AM - The Conflict Question
The transcript from 2:08 a.m. shows a call from James Liang, a matrimonial attorney in Boston. His intake is solid. Matrimonial dispute, custody angle, out-of-state assets. The conflict flag is red.
I read the agent's note: "Caller mentions retaining an accountant named David Chen to analyze the other party's financials. The firm's database shows a David Chen retained for expert analysis in Kozlov v. Kozlov, 2019."
This is not a conflict. This is a coincidence. But the agent flagged it red because it matched the name in the database. I need to fix this.
I open a Slack message to my own channel - I use it as a notebook, to audit the agent's behavior. "Check: name-matching false positive in database conflict check. David Chen is a common name in Massachusetts. Need to add company/jurisdiction checks to avoid flagging unrelated experts." I add it to my Linear board as a bug ticket: "Conflict flagging logic: too sensitive to name matches, needs filtering."
Then I open Gmail and write to James. "Hi James - the overnight flag was a false positive on an expert's name. You're clear. I'm attaching the draft retainer. Looks like you'll want to add the asset-discovery timeline to section 4. Callback window you requested is 8 a.m. Wednesday. The intake is on your desk this morning. Let me know if you want any changes to the engagement terms before signing."
I think about this for a second. The agent took the call, asked the right questions, ran the checks, drafted the retainer, and prepared the transcript. I spent fifteen minutes this morning fixing one bad flag and clearing three cases. The overnight system generated six hundred dollars in monthly recurring revenue while I slept.
10:47 AM - Admin Review
I check the dashboard metrics. This is the part that still feels strange - I'm watching a system take orders while I sleep.
Week-to-date signups: 12. That's two more since yesterday morning. The sales motion is still outbound cold email to solo practitioners, mostly from LinkedIn list-scraping. One of last week's demos closed yesterday - Carol Reyes at Reyes Family Practice in San Antonio. Carol signed a $280 monthly contract. That's recurring revenue. Stripe shows $280 coming in every month from Carol unless she cancels.
Today's new signups are smaller. A two-person firm in Vermont. A solo practitioner in Florida who took the trial plan at $99 a month. That's a short trial, but if they see the revenue from one successful overnight intake, they often upgrade.
I open my Slack workspace. There's a notification from the escalation queue: "Call failed to connect at 7:15 a.m. to intended recipient. Customer contact number appears to be invalid." That's from one of the Massachusetts firms. I find the case record and note that the callback number has a 617 area code, and the timestamp shows it was supposed to ring through at 8 a.m. The agent tried. The client didn't answer. It happens. I manually email the attorney with the intake and ask her to circle back with her client directly.
One of yesterday's customers - a small criminal defense firm in Denver - sent a support email asking about billing. They were charged $150 but they signed up for the $99 tier. Looking at their contract, they triggered the overage charge because they had six intakes in the first week and only the first four were included in the monthly fee. This happens. I open Gmail and send them a note: "Thanks for catching this. The initial volume was higher than expected, which triggered the per-intake overage. If you want to switch to unlimited at $199 a month, I can backdate that and zero out the overage. Let me know what works for you." Sometimes it's faster for me to absorb a hundred bucks to keep someone happy.
12:30 PM - Lunch and the Pipeline Check
I eat a sandwich at my desk and open the pipeline sheet. Twenty-seven firms are in the active demo phase. That's two weeks of demos at the target rate of 1.5 per day. Not all will convert. Typical close rate is around 12 percent, so I should expect three or four to sign. The ones who are already taking calls and watching the thing work are the ones who close.
I make a call. Actually, I decide to make a call. This is the part that I wasn't expecting when I bought this. The system works. But someone has to follow up. I dial Jennifer Park at Park Legal Group in Chicago. She signed up for the trial plan six days ago.
"Hi Jennifer. I'm following up on how the system has been treating you. Have the overnight calls come through yet?"
"Yes, two calls so far. One was a solid lead. The other one... I'm not sure about. The intake form was good. I just haven't had time to sort through the conflict check on the second one."
This is the sticking point. The system works. But solo practitioners are busy. They need a little push to understand the value.
"The conflict check is automated, so if it flagged green, you're clear. If it flagged yellow or red, I review it personally before it hits your desk. You won't see a red flag unless it's worth your time. What's the second intake on?"
"Family law. Custody dispute."
"Any red flag on the conflict check?"
"No, it came through green."
"Then you're good. It's real lead. You can take it or pass, but there's no conflict issue. I'd take it if I were you."
She laughs a little. "Okay. That helps. Yeah, I'll take it. Actually, I think I need to move to the unlimited plan. I'm getting more calls than I thought."
I close the loop in Stripe and update her subscription. One hundred ninety-nine a month, recurring. That's her choice, but it's also the choice I wanted her to make.
2:08 PM - A Sticky Edge Case
A customer emails me directly. Robert Eisner, a solo practitioner in Nashville. He's on the $99 plan. He has a question about the retainer template and whether Counsel can modify it to match his specific fee arrangement.
The agent system can work with templates, but custom modifications require manual intervention. I've been handling this myself because Eisner is new and the request seems reasonable.
I open his account. He wants to add a clause about a reduced flat fee for uncontested divorces. That's standard language. I don't need to ask him to code it or hire a developer. I spend twenty minutes finding a clean version of the clause in another customer's retainer, adapting it to his firm, and uploading it as his new default template.
I send Robert an email: "Updated your template. This should work for your uncontested flat-fee cases. Just let me know if you want any other tweaks."
This is the unsexy part. The AI does the heavy lifting. I do the human work. Twenty minutes of my time probably saved Robert three hours of contract drafting. He probably doesn't realize I did it manually. He thinks the system is magical. Maybe it is, a little.
4:30 PM - Customer Churn and a Win
A notification in Slack: Cancellation request from the Vermont firm I signed two days ago. I check their account. They had one intake, attempted to call the client back, and the client told them the case had already been resolved elsewhere.
That's not a failure of the system. That's how intake works sometimes. But I could have lost them at this moment. I send a quick email: "I saw you had one call through the system. These things happen with intake - not every caller ends up being a real lead. But the system will keep working for you overnight while you sleep. Before you cancel, would you be open to trying one more week? I'll waive the plan fee for next week and you can see if the call volume gets better."
That's twenty-five bucks out of my pocket. But losing a customer after two days looks bad in the metrics, and I'm still learning which segments of the market value this most. Vermont stays for one more week.
Meanwhile, Carol Reyes - the San Antonio family law attorney who signed two weeks ago - sends me a Slack message. "Hey, just want to say thanks. The system caught a call last night from someone looking for a contested custody modification. I landed the case. It's a $4,200 engagement. Your system paid for itself in one week."
I screenshot that message and save it. That's the moment that made the whole thing worth it. Not the revenue number. The fact that someone is actually using this and it's working.
6:15 PM - Wrap
I close the dashboard. The intake system is running fine. Nothing is on fire. No critical bugs. The false-positive conflict flag will take me an hour to fix tomorrow. Three customers are active. Revenue this month will be around $2,400 if the Vermont firm stays and the Florida firm upgrades.
That's not scale yet. But it's real. Real customers. Real work. Real revenue.
I close the laptop and think about what's working and what needs to change. The demo-to-close rate is too low. I need to either improve the product experience so more customers stick after the trial, or I need to spend more time on the phone with prospects explaining what they should be looking for in the system's output.
The conflict-flagging logic is too aggressive. I'll fix that tomorrow.
The manual work - the email follow-ups, the template customizations, the customer success checks - that's not scaling. But right now, that's what keeps people. The system does ninety percent of the work. I do ten percent. I'm okay with that math for now.
Tuesday is done. I did real work. The system did real work. Counsel took maybe four calls overnight, caught conflicts that mattered, drafted retainers that made it to someone's desk by 7 a.m., and generated revenue that I can see in Stripe.
That's the day. No magic. Just leverage.