PER: @wheattoast11 over here
“*… The gold is in what traces you’re teaching the model and how you can “connect the dots” for the model on what it has learned. LLMs “know” far more than any human possibly could. However it’s our job to teach them to understand what they know, and that happens through training, either in context through each inference pass for a single conversation (ie no permanence), or through training the model to enrich its ability to do this intuitively (and memory persistence)…”
I’ve been in my bubble about Symbol Based Holographic Context sharing for Agentic Computing and Orchestrationg.
The concept is zeitgeist-ing heavily right now (such as agentic SWARMS, and we are all circling the same singularity - and as its been said “The singularity is very near, we just dont know what side of it we are on yet”
I was reading about watermarking postgres
- Using watermarks to coordinate change data capture in Postgres*
And with looking at how SpecStory was reading from the SQLite, I wanted to pipe directly to postgres my composer history and the code it spits out.
I then wanted all of my YOLO_Agent’s Terminal Commands “Y’mands”, as they’ be known. to be captured and tracked - I direct my Yolos to wrap all commands into a .ps1 and then execute the .ps1. The .ps1 is to have a header, a description. and invocation instructions.
I even went on to have its readme.rmd entry…
Then I wanted all my .ps1s posted to a catalogue in postgres and a workspace directive to point the agents to the postgres table with .env db creds, and grab context from the composer table - and I can tell the agent which composer session to begin interacting with.
Further I wanted the agents to register a file_locking system and thus it was to be built upon github repo that the agetns use gh cli to interact with as collaborators in the repo…
And ive really been interested, as I interact with the agents more and more with a clear intent on where I want to go, but I have to walk through my inner dialogue as it manifest in a fractal way as I interact with the agetn – meaning:
I have an Vision that has a clear Feeling that is pulling me… but I dont yet have the vocabulary to add resolution to the vision, and as I interact with the Agents, the vision become more clear, the feeling stronger - but it fractally scope_kreeps into all sorts of other what_ifs and oh_and_i_need_this
CursorFocus by @RenjiYuusei aligns with other focus and ignore issues… we have been talking about.
(I am also having the agents check out their Role|Archetype|Domain_of_Expertise from an assignment when spawned) - the agents are spawned by an orchetrator agent…
So, with that preamble:
TL;DR Agentic quantum state holographic geometry nodes…
Now with 100% more TCP!
BECAUSE – once you understand how to get the model to create a comlpex, symbol based Context_Lattice - but as a Ruleset and it can use condensed symbols for vastly complex further actions, which can be paralleized, compartmentalized, sub_computed(sub contracted)…
It makes fun relationship deep dives like this far more powerful:
(Reasearch for a game I am developing, called ‘The Oligarchs’)
