Chief Architect currently working on Humanistic Practices and who loves: Computer Languages, Math, Music, Archeology, History, Literature and teaching.
The Pattern Language from Christopher Alexander is the real thought process. So LLMs with command of language will understand and greatly benefit from Design Pattern guidance. This is going to be fun.
Design Patterns never reached their potential because everyone fixated on the solutions (object diagrams) and lost track of how the solutions had come about and their applicability.
I am finding the IntelliJ AI Assistant with Junie to be very useful in coding and debugging since it has access to the code. Since Junie can run an agent orchestration harness I can refine my 9 practices. In addition the session / artifact memory will be able to remember context.
@Emotion78687 A truly Hot Band with Emmy leading it with sublime vocals and having the band keying off of her rhythm guitar. It just doesn't get any better than this.
@HowToAI_ You only get knowledge collapse when you let AI do your thinking for you. However you gain capabilities when you pursue knowledge augmentation like curating your favorite areas of study with NotebookLM. So I find the whole premise of the paper to be utter nonsense.
LLMs only reach their potential when provided with solid up front information. For me its Learn:Domain, Build:Construct and Share:Deliver with NotebookLM curating best practices. Otherwise it is just vibe coding with no memory of what has been accomplished.
@tlberglund Models only reach their potential when provided with solid up front information. For me its Learn:Domain, Build:Construct and Share:Deliver with NotebookLM curating best practices. Otherwise it is just vibe coding with no memory of what has been accomplished.
In the past developers glazed over detailed requirements, but now LLMs eat them up and can automate traceability from features to code to acceptance tests.
First Google Anti Gravity and now IntelliJ AI Assistant are playing a shell game about "Usage Statistics" which I need to provision my paid accounts, especially when the AIs gallup out of the barn, do hundreds of things I don't want that consume tokens. I just want to do business
So I am now happy to the see embrace of Kanban with agents, especially since you pull work from queues. In addition agents work well with detailed requirements for understanding the business along with your personal preferences for guidance.
Well my Information Agent Orchestration Practices are coming together for test runs this week. The irony is that AI affords me to indulge in by age old biases of synchronization Bun:SQLite that is good for review and top down waterfall requirements that make the best guardrails.
Long ago I kicked Agile to the curb and replaced it with an iterative Adapt / Kanban approach, mainly because Agile sprints were a demoralizing treadmill and user stories were a train wreck compared to solid requirements.
@techgirl1908@goose_oss What I like most about @goose_oss is that everything configured perfectly right out of the starting gate, especially with Gemma. This along with easy access to skills and /commands made a deep positive impression on me.
Tom Flaherty
Morons like Trump and Magas can't stand being called out for blatantly ridiculous and stupid behavior, which provides endless opportunities for satire. Magas are incapable of owning up to their foolishness and recognizing the basic sensibilities of their audience.
Magas have to resort to media control which destroys brands like CBS. I swear finding material to mess with Magas is easy. The hard part is dealing with illegal repercussions like the FCC.