Entrepreneur | Rocket Scientist | Fndr
@ctlst_tech
| Fndr AI orchestration tool for boosting doers
@apply_ai_app
| Father of 👧🏼👧🏼👧🏼 | My journey & Memes
Running fully autonomous news digest about AI from X:
- collecting data from several account categories;
- filtering out non-AI news;
- compiling digest on the schedule;
- sending results to telegram channel and Reddit
#ai
#chatgpt
#generatieveai
#automation
#ainews
#autogpt
@garrytan
Police did not even come to my call when I got my laptop stolen from the co-working space on Bryant St. in 2019.
Gone pitching for 5 minutes, boom, it's gone!
I had to come to the office in person to file a report. I sent a picture with direct evidence from co-working CCTV.
"Dear Santa, for Christmas, I wish to embrace simplicity in my work and stop overengineering. May I find the balance between creativity and efficiency."
#buildinginpublic
You can look forever at three things:
1. fire burning;
2. water flowing;
3. Other people working;
+ Your app's internal dashboard
#buildinpublic
#indiehacker
A huge amount of knowledge is stored in countless working chats, random notes, scattered all over the place. Now, it can be reused and converted into actionable and reusable items x1000 faster than it was possible before.
Just an admirable fact. Best times to build.
This is me bursting into the new week with the determination to finish up semantic search-based features and generalized synchronous bot’s interactions 😸
Founders
#letsconnect
!
After a week of talking to b2b users.
Leadership and discipline cannot be substituted by tech and automation;
But tech can still multiply the bandwidth of the leaders.
The picture below is a RAG setup example for multiple data sources in the apply-ai platform. RAG+LLM is a
This is me bursting into the new week with the determination to finish up semantic search-based features and generalized synchronous bot’s interactions 😸
Founders
#letsconnect
!
When we showed our first users a no-code canvas, it was quite confusing for most of them. So we simplified the initial setting of the agent template to the simple form UIs.
Work in progress.
#buildinginpublic
Life as the matter that replicates itself. It stands not on carbon and water but on a feedback loop with negative feedback (-), which keeps its vital parameters in a dynamic equilibrium because static equilibrium is death.
And then the set of nested feedback loops is what makes
Hey builders,
What would you do with a machine that is able to consume and digest information for you?
I use it to wrap AI buzz into a 1-page summary daily.
#buildinpublic
If you are LLM enabled startup or team, I can save you months of building with this.
It is a RAG infrastructure-building tool for indexing and searching.
Just click blocks, drag and drop data or data sources,
connect to your app via the interface you like.
And start querying.
How often have you heard “reinventing the bicycle” in a non-supportive context? [build what do exist already]
#tech
In the meantime incremental innovation within anything is magical. Look at the pictures below.
This is the story I genuinely admire. Modern advanced bicycle is
In our small team of two, we've streamlined our app's self-testing process:
1. I record a voice message saying "run regression test."
2. This message is sent and then received through the messenger platforms we support in the app.
3. It undergoes tagging, rounding, and other
We were testing Telegram's chat knowledge retrieval app in the public group.
Some time ago, I asked users how they want to improve app's behavior and how we could avoid training data contamination.
Instead of searching for that thread and messages, I asked the app itself.
I
My personal benchmark for testing pictures/video generative AI is
"yellow rubber prank chicken exiting hyperspace"
The picture below is the best I've got; as you may see it is far from perfect, but improved dramatically. Kudos to DALLE-E
When survival chance is a life's objective function. The following of curiosity feels like the top-layer optimization method for it at the moment of Earth's history.
Experimenting with UX around gen-AI
1. Query my transcribed voice memos DB in Notion
2. Evolving content
3. Generate landing page YAML config
What do you think?
+ laser precision context control
+ initial context reuse with less effort
+ less ctrl+c ctrl+v
#buildinpublic
Interesting point.
Fundamental things are harder to master without a tutor.
Because the reality check feedback loop is expensive.
In CS, we basically iterate any skill against the computer. The iteration cycle cost drops down to minutes. Skill and knowledge are polished much
I changed majors my sophomore year from Computer Science to Math because my Electrical Engineering course was too hard.
Sometimes I regret that but it turned out to not mean much in the long run.
Do you guys know any existing private audio podcasts platform?
Where I can publish audio tracks via API and have a of the shelf mobile app to listen for these podcasts?
What public content at scale would you like to query this way?
Keep playing with no-code configurable RAG and knowledge retrieval and here is one of a public demo (search . apply-ai . app)
The demo is a semantic search over a bunch of public telegram channels. While telegram is
They are evolving chat based interface to its limits. Idea is to become a hub.
But chat interface is already quite limited, single thread request-response is limited.
We need an interface to stack ideas using quick gestures
Most exciting time to live in yet another transition to the plow
That’s deserves to be in e/acc hall of foundational memes;
chamber of weights and memeausers;
hydgromeme clocking standard;
meme pulsars constellations maps;
To safely and responsively navigate-memegate the
Project
#2
: LLM Visualization
So I created a web-page to visualize a small LLM, of the sort that's behind ChatGPT. Rendered in 3D, it shows all the steps to run a single token inference. (link in bio)
@pronounced_kyle
The only radiation effect which is relevant to nuke is SEU and even civil aircraft avionics (DAL A) has to handle that effect by their architecture (not because of nuke but because of cosmic rays at high altitude). So do the LEO satellites with good design.
The absorbed