A yes from a university, and four apps rehomed
Week four, and the midpoint of the two-week deadlines I set last Sunday. One conversation set up the next month of client work, a lot of infrastructure moved behind the scenes, and it was a really busy, and satisfying, week on the personal side!
The headline
The best meeting of the month happened on Wednesday.
I had a first real conversation with the technology leader who now owns AI for a university I advise. Forty-two minutes, and I left committed to running an envisioning session for a small group of his people: not the skeptics, not the early adopters, the messy middle. He was candid about where they are. A campus full of tools, a lot of people using AI as a search engine, and a culture that has done things the same way for decades. That last part is the real work, and it is exactly the work I want.
Right after the call I recorded a memo to myself: good discussion, now lock down the timeline. The note that covers actions, owners, and the window is in draft, and sending it is this week's first job.
Shipped
- A yes from a university. An agreed session, an agreed shape (about an hour, three scenarios, then a reset with him on what landed), and three things he is sending me so the session fits his campus.
- Four of my own apps moved off Replit and onto Railway, with logins and secrets set up properly on the way. Agents did most of the lifting; I made the calls and checked the results. It is the same move I expect to recommend to clients: get off the platform that was easy to start on and onto one you can actually run.
- Penny handled the travel receipts. I forwarded them, she filed them, asked me what she could not tell from a receipt, and recorded my answers. Minutes, not an evening.
- Issue 3 went up on Tuesday. One command, as designed - posting progress, with very little effort.
- Walking normally again. Stairs included. That one was on last week's list, so it counts - glad I ran the marathon, happy that stairs aren't an obstacle this week.
Started
- The session itself. Tailoring it to the tools the campus already has, so the examples land on what people can open tomorrow instead of what I happen to use.
- The call with two builders is on the calendar.
Did not move
These are on the two-week clock, which closes next Sunday.
- The university follow-up. Drafted, going out first.
- Two outreach notes. Not sent yet.
- The rename. Still open.
- The check-in with the team I ran a session for. Moved again on the calendar.
- Penny's hosted spec. Not reviewed yet.
- Funnel, the client records app, and the research desk. No changes. All three work as they are, and the infrastructure week came first.
Stopped
- Running my own apps on a platform I could not fully see into. Four of them now live where I control the deploys, the logs, and the secrets.
What I learned
- Write down the risk while it is fresh. My memo after the call said "no timeline" before I could talk myself into thinking it was all settled. It turned a good meeting into a to-do with an owner. I would give any client the same advice: the ten minutes after a call are worth more than the hour before it.
- Start with the messy middle. The skeptics will not be moved by a demo, and the early adopters do not need one. The people in between are where adoption actually happens, and they are who the session is for.
- Easy to start is not the same as easy to run. The platform that got my apps live fastest was not the one I wanted to own them on. Moving them was a few days of focused work with agents doing the heavy parts. That is a good sign for anyone sitting on prototypes they built quickly and now depend on.
Next week
Send the university follow-up with dates and owners in it. Then close out the two-week list by Sunday: the two outreach notes, the rename, a time for the team check-in, the builders' call itself, and a plain answer on Penny's hosted spec.