spencepd retweeted
This auto-eval loop is interesting.
For my own agents the key takeaways for me are:
1. have clear evals
No eval → no improvement loop.
Define:
* completeness score
* correctness checks (did tool calls succeed)
* structure validation (output matches schema)
* decision usefulness (business logic)
2. Log full traces (not just outputs)
You want: tool calls,.decisions, failures
dead ends. Not for debugging
for learning and iteration. I.e. want:
* tool call inputs/outputs
* reasoning steps (even partial)
* failures (timeouts, bad data, wrong tool)
* retries / dead ends – at Alpharetta, GA
spencepd retweeted
Check out the EyeOnMajors Fantasy Golf Cut-Making Leaderboard for 2025!
eyeonmajors.com/patron/leade…
spencepd retweeted
Make The Cut and Win Free Fantasy Group Pool
Compete in the Players Championship fantasy golf pool on EyeOnMajors -- If you just make the cut, you earn a free major with EyeOnMajors pool for the Masters. Give it a try! eyeonmajors.com (1/5)
spencepd retweeted
Arnold Palmer’s driver off the deck at Bay Hill
eyeonmajors.com/golf/news/ar…
"When Arnold Palmer famously hit a driver off the deck, he was 74 years old and the shot occurred at the 2004 Arnold Palmer Invitational; the exact score for that day is not widely reported, but the highlight is remembered for his daring decision to use a driver from a difficult lie on the 18th hole, showcasing his aggressive playing style even at an advanced age."
spencepd retweeted
We're growing. If you are interested in exploring how we could work together, let's talk. #softwaredevelopment #humanfactors #lean #mobile
spencepd retweeted
As we wrap-up 2017, it's exciting to reflect on achievements. Thanks for the list, @Redox. In 2018, let's add categories for #EaseofUse and #PhysicianHappiness. Our nomination is: #heathii. Cheers! bit.ly/2AaXfWw
spencepd retweeted
Deep learning #AI for pathology slides: best study published to date, shows at least equivalence for machine accuracy dx #cancer mets cf pathologists @JAMA_current jamanetwork.com/journals/jam…
spencepd retweeted
Human Processor Model: bit.ly/1gyvaHu - cognitive modeling method to calculate how long it takes to perform a task.
spencepd retweeted
The Adapter Pattern in Rails: monkeyandcrow.com/blog/readi… (A look at a common design pattern implemented by Rails.)
spencepd retweeted
Here's our latest cover, featuring a story on the bourbon boom. What's your favorite bourbon? #FortuneBourbon
spencepd retweeted
Now you can slice/dice Google image searches by usage rights under "Search tools. Thanks to @lessig for the request!
spencepd retweeted
.@designoutloud this one is for you. Cross-cultural design/semantics – at Ohio, USA
spencepd retweeted
Why Mobile First?
youtube.com/watch?v=2m-PqCnd…
2.5min video interview with me by @intel
spencepd retweeted
8 things the Marines aren't telling the Navy i.imgur.com/TMUFIPC.jpg
spencepd retweeted
When you step out of an elevator you always take a step in the wrong direction.
#Fact
spencepd retweeted
Tomorrow's the big day. @ChefCMcKenna's favorite sauce - Bourbon Barrel Aged Carolina Gold - will be available for a limited time.
spencepd retweeted
These guys have been making a nice living by #podcasting for years. Now they wrote this (& I helped): UltimatePodcastingGuide.com
spencepd retweeted
A solid edit of our show in Portugal last weekend. I forgot how many people were there! onsk8.com/videos/moche-tony-…






























