Three big asks come in through various channels, all from the same client. One of them was actually me, asking the COO if he had more AI automations for me to take on.
Wouldn't I be breaking my own rule? One task at a time, or nothing gets done. I complete one task, and then move on to the next. This is the DesignJoy model I lifted.
Each of these tasks is multi-week. The one-task-at-a-time rule has been the AI operating cadence I write about most weeks. So far it has worked. Why am I purposely breaking this rule?
Ask one: research two enterprise market-data and research vendors. Integration, cost, contract, plus how either would plug into the AI research assistant we are building on the side. Multi-week, maybe. Not even sure how to approach this one because it will require me to peel off other projects to inspect the code that has been written to make the right call. Perhaps a task for Jan. Need to think that one over.
Ask two: take the AI Financial Controller that goes live for the company on June 1 and extend the same playbook into K-1 generation across the partner entities. Different domain. Same plumbing. Multi-week, but also in the overall scope.
Ask three: automate the investor deck assembly for the next round. Charts from the live ledger, narrative from the quarterly read, deck template kept consistent across raises. Multi-week, but I'm willingly taking this on as well.
Three big asks. One me.
Each one is the highest-priority piece of work for the person who asked.
I open the one-task-at-a-time rule I have been writing about for the last fifty-five days. The version I have been telling people needs an edit.
---
The one-task rule says you pick the one thing that has to ship this week, or the next few days, you ship it, and the rest of the work waits. Fine rule. Not the full description of what an actual working day looks like.
What an actual day looks like, for the last six weeks, is two cycles. I write at 6am. I send a question to the operator, the CFO who has to confirm a transaction route, or make some tests in the sandbox, or send me more data to back-test the build for the AI Controller we are building to bring down their hours. The reply lands at 11, sometimes even later. That's fine. I understand other work is being done, but it means I'm basically left waiting for that info to move the task forward, and I'm blocked. When the info comes in, through Trello as per the DesignJoy model, I send the next question in under an hour. The reply then sits, and I'm blocked again. Two cycles. Not eight.
Between those cycles I am waiting. Not in a small way. Half the working day, sometimes more.
The one-task rule is a useful idea for a single seven-day week. It is not the right rule for what I do at 10am between the two emails. At 10am I am not blocked on the one task. I am blocked on the operator. The operator does not get faster because I sit there hitting refresh on Outlook.
I have been pretending not to notice this. The rule works for what I commit to ship this week. It does not answer what I do during the gaps.
So I keep a queue.
Three or four pieces of research, draft writing, code review, anything the operator does not need to weigh in on. Vendor research fits in the queue. The K-1 build does not fit yet. It needs the operator's voice on what the partners actually want from the document, and that conversation has not happened. Deck automation fits, partly. Investor deck templates and chart styles do not require operator input. So, I reach out to the COO and set up a call with the head of investor relations.
What goes in the gap queue is the work that does not need the person I am blocked on. Everything else waits, and I'm ok with that. My main priority is to do such high-quality work that this client becomes our use case, gives recommendations to their network, and we expand the reach with the overall ONE Thing strategy I wrote about yesterday.
Three asks land in one morning. Two are from people I want to say yes to as fast as possible: the founder building the research assistant, the partner who runs the place. I want the yes. I say yes. I say yes to all three. I do not name the cadence. The senders read "yes" and assume "starting now." Three Fridays later, two of the three have not been touched and the third is half-done.
What I do instead is queue the yes. Yes is me starting now, then waiting on you.
The queued-yes is uncomfortable in the moment. It reads as slower than what the sender wants to hear. It is also the only honest version when one person is doing all the work.
What I do is add a Blocked column in Trello, and list why I am blocked. Then at least I have a workflow rule that shows the status, why I am blocked, and why I am taking on more than one task at a time.
At some point the queue tells you to hire.
I did that this week. My CTO Jan looked over the AI Financial Controller build, as I wrote yesterday. The reason was as much queue overflow as anything else. The Controller go-live is June 1. The three new asks landed Thursday morning. If I am the one writing every line of Phase 1, am I really the right person to be doing the work, or is it time to bring in help? The right help.
Jan takes Phase 1 to go-live. I take the three new asks into the queue and start working them in week-one, week-three, and week-four cadence behind the BIP writing and the prospecting calls, meanwhile we really only have revenues just for me to work full time.
The test for whether Jan in turn needs help is whether his queue overflows three weeks from now, then tied back to how can we afford to do all of this on fixed monthly revenues that haven't grown.
Balancing one task at a time is what I am working on now. A queue for the gap windows. Work the operator is not blocking. A hire when the queue overflows past what one person can carry in the gap windows, and that's not predictable.
Classic startup stories, live in real time.
I will probably be running a different cadence by July. The BIP exists so that when it does, it goes wrong in public, in front of the people who would otherwise read a tighter version of the same story two years late.
Day 56 of 365.
Monthly Revenues $11,800 | Clients 2 | Prospects (AI marketing employee live in 6 days) | Team: Me + Jan (CTO)