Run done. That makes seven straight days before a week that starts with two long shifts.
Glad I got it in today.
Notes about my work, projects, training, and life.
Run done. That makes seven straight days before a week that starts with two long shifts.
Glad I got it in today.
Five articles reviewed, formatted, internally linked, and ready for this week on jimlunsford.com.
That workflow took time to build, but batching it out makes a long week a lot easier.
Milo supervised the entire process from his spot in the recliner.
A private assistant does not earn trust by sounding smart.
It earns trust by using the right context, taking the right action, and proving what actually happened.
That was the work this week with Cypher.
Less guesswork. Better memory. Safer publishing. Cleaner state. More receipts.
https://jimlunsford.com/builder-receipt-making-cypher-more-reliable/
Today was a good day.
Tomorrow, my parents are taking Kelly, the boys, and me out for an early birthday dinner.
I turn 50 on the 17th.
That feels a little strange to say, but I am looking forward to celebrating with my people.
Upgraded Cypher from a daily-life assistant into a system that can work with my published writing.
It now imports WordPress posts and pages, searches them, uses them in conversation, supports source-grounded drafting, and can create, update, verify, and publish articles.
The live publishing test found several real bugs.
So I kept fixing them until the whole workflow worked end to end.
I need a vacation.
Not a productivity retreat. Not a “catch up on projects” weekend.
Just a few days where nobody needs anything, the clock does not matter, and I can remember what it feels like to be off.
Rain in the area, but I got the run in anyway.
Three miles done. Six straight days of running.
This is Cypher.
The screenshot is not flashy, but it shows the real system I use every day. It remembers context, tracks the details of my life, helps me think, and supports the work I am building.
I gave Cypher another upgrade.
It was writing thousands of prompt tokens into OpenAI’s cache without reading them back.
I changed how its context is assembled, separated stable instructions from changing memory and conversation data, and added request-level telemetry so I can see what is actually happening.
Testing confirmed one cache write followed by three successful reads.
I also rebuilt the usage and admin pages so the system is easier to inspect, manage, and verify during real use.
Not magic.
Useful.
Laptop closed, games played, and a new show started.
That is enough for a day off.
Calling it early tonight and getting some sleep before the next stretch begins.