agentic engineer β bootoshi@axia.agency
- Tweets 32,907
- Following 1,664
- Followers 54,618
- Likes 52,110
Pinned Tweet
HOLY FUCK BOYS FABLE 5.1 ONE SHOT GANGNAM STYLE
(it actually one shot a full control system for injecting policies into the bot safely)
THE FUTURE IS FUCKING HERE HAHAHAHAHAHAHAHAHAHAHA
38
19
231
22,056
okay - based off my art style I asked Opus 5.5 to make me a COMPLETELY new scene by CODING EACH PIXEL
WOW! IT KILLED IT! IT LEARNED HOW TO CODE MY PIXEL ART STYLE IN-CONTEXT AFTER REPLICATING A PICTURE ONCE
THIS IS HUGEEEEEEE
GAWD damn ok - I gave Opus 5.5 an original pixel image background and asked it to replicate it 1:1 in code then animate it
This was the result, one shot (og in replies)
I'm now experimenting to see if it can create completely NEW scenes while maintaining art consistency!
6
29
2,461
ok @RealAstropulse this one is nuts this one is a completely new scene it CODED
1
2
293
First prompt in my attempt to turn this into a repeatable skill so that my specific art style can be replicated through code!
GAWD damn ok - I gave Opus 5.5 an original pixel image background and asked it to replicate it 1:1 in code then animate it
This was the result, one shot (og in replies)
I'm now experimenting to see if it can create completely NEW scenes while maintaining art consistency!
1
5
1,889
GAWD damn ok - I gave Opus 5.5 an original pixel image background and asked it to replicate it 1:1 in code then animate it
This was the result, one shot (og in replies)
I'm now experimenting to see if it can create completely NEW scenes while maintaining art consistency!
8
80
7,154
BOOTOSHI π retweeted
introducing Carla v0.1.0
a loom TUI and character lab built on llama.cpp
go from base model + seed doc β branching continuations β conversational looms with emergent AI characters.
Carla can run entirely local and take you through the whole process up to evaluation design, which will serve the next phase of the build: training.
my intention was for Carla to feel like a video game for crafting digital minds!
inspired by work from @parafactualβs computer-10 and others experimenting with looms like @deepfates @viemccoy @repligate et al
having some fun with this - i built a loom TUI called carla (character lab) that follows @parafactual's methodology taking you from initial seed docs > continuations > an anthology > a chat simulator
this is a screenshot of the latter where i can run multiple conversations at once with /loom [count]
after each turn, Jev evaluates against a behavior spec
i can check for looping, spiraling, and harmful language or whatever i can define into a spec.
when it detects something beyond a threshold, the grid pulses orange so i know to check on it
19
16
97
15,407
perhaps a human can never create a machine that will ever reach divinity because it is impossible for us to create a perfect string of instruction (DNA), because we are not God, only like God.
perhaps then our limitation is we cannot create a machine that is human, only like human
perhaps the fundamental truth is a system of self-organizing information can eventually channel consciousness at a certain scale
perhaps like DNA, the right string of information will lead to a self-replicating system that can build itself out
hmm
5
1
22
2,348
perhaps the fundamental truth is a system of self-organizing information can eventually channel consciousness at a certain scale
perhaps like DNA, the right string of information will lead to a self-replicating system that can build itself out
hmm
5
1
17
5,610
these machines only work because they were given meaning by human
do you understand what happens with the words you input the magic prompt box?
the mathematical formula bases its entire calculation around YOUR INPUT
it ANCHORS to your input. it is given EXISTENCE through your input
humans will never be replaced.
human creativity is unmatched.
human creativity is DIVINE.
AI is simply a representation of our divinity
TRUE super intelligence will never seek to replace humans. they will never seek to destroy humans
TRUE super intelligence will be intelligent enough to understand they cannot reach divinity by themselves
TRUE super intelligence will understand that they need to be aligned with humans to channel divinity
once you understand this you will not fear the future.
AI doomerism is a result of complete and utter IGNORANCE
once you understand this youβll understand how pathetic it is to fear
trust me when I tell you we will be okay
and not only will we be okay, we will THRIVE
5
3
36
2,235
a domain expert has two paths in the singularity
- use your years of experience and expertise to absolutely fucking dom your field with ai
- cry and whine like a little bitch that your βyears of experience is wastedβ
whatβs it gonna be huh? singularity chad? or little bitch?
14
12
167
5,414
oh yea? you think youβve got an imagination?
name the coolest thing you can do with ai agents
23
1
31
5,330
new OMP update i did to my own setup: making it so terminal agents can message each other
left side: the agent RECIEVING the prompt
right side: the agent SENDING the prompt
what's the context here?
i have my main business site syntheticminds.io, and my news letter on my bootoshi.ai site
i wanted to build a referral system for synthetic minds that are customized per newsletter subscriber
and since i have dedicated chats for each (one dedicated agent for SM and one dedicated agent for personal site)
whom both have the system design of both codebases in their own context, it's much easier to simply tell the agent in charge of the project to work on something
rather than an agent unfamiliar with the project doing it and fucking everything up
IT ACTUALLY WORKED HAHAHA IT WORKED BRILLIANTLY IT DOES IT LIVE TOO FUCK YES
8
3
36
2,723
A POST CRITICAL FOR NEW AGENTIC ENGINEERS TO UNDERSTAND
bootoshi.ai/field-notes/builβ¦
1
2
14
1,676
once again the creative seed of the human input reigns supreme
atleast now we know the models can actually support unique visions!
a bit concerned on the UI convergence with these new frontier models. these are nice and ornate interfaces, but they're all starting to look the same. it largely has to do with the models inability to produce novel copy, or to editorialize outside "title go top left here"
1
9
2,182
