4339
Mo
@meb #4339
Designing and building AI systems
https://agents.gladio.ai
936 Follower 620 Following
Question; How on earth do you get token usage stats for Gemini API?
UX feels way more complicated than it needs to be
UX feels way more complicated than it needs to be
Strong foundations are important
Primer on keeping your token costs down
https://www.reddit.com/r/OpenAI/comments/1kiglaa/spent_9400000000_openai_tokens_in_april_here_is/
https://www.reddit.com/r/OpenAI/comments/1kiglaa/spent_9400000000_openai_tokens_in_april_here_is/
Code syntax == metaphor for the world
1349
avi
@avichalp.eth·4 days ago
this is going to be important even if most code is written by LLMs.
you still want to understand your codebase. sloppy naming makes that hard
you still want to understand your codebase. sloppy naming makes that hard
Something something evals
13596
hellno the optimist
@hellno.eth·4 days ago
probably mid-curving it, but I love promptfoo
Bill Gates would be proud
380950
silly goose
@lovejoy·5 days ago
Pope election smoke system in case you were also wondering how it works
My lime stats over a year in and out of Berlin. e-scooters in this city are so fun
Patched my 5 year old JBL Flip 5 speaker firmware from v0.3 to v0.5.6 today. Power on/off bugs and pairing issues gone.
PSA to keep your dependencies up to date.
PSA to keep your dependencies up to date.
It's kind of crazy how LLMs as a class of AI models have lifted our society.
What other emerging model types do you see holding promise?
What other emerging model types do you see holding promise?
In the midst of this whole drama, clanker comes to totally misread the room lol
3 values I took from my walk with ChatGPT in Berlin’s military cemetery with some of greatest heroes of its long history, that can apply to building great agentic systems:
Clarity: Understand reality of world around you. Know what the next moves and plan are.
Discipline: Have your things in order you can mobilise any capability when needed.
Courage: bias for forward motion; remove friction, make it easy to press send.
Clarity: Understand reality of world around you. Know what the next moves and plan are.
Discipline: Have your things in order you can mobilise any capability when needed.
Courage: bias for forward motion; remove friction, make it easy to press send.
Just published first NFT of collection I'll add to this summer
Based Berin #01 - Spaceshark the piano player, Rudolph the red-nosed reindeer, and stick man
I was in a queue waiting to enter a party in the forest, when these ladies offered me to contribute something to the drawing they were doing together (I added stickman, but the hat and suitcase were their idea)
Based Berin #01 - Spaceshark the piano player, Rudolph the red-nosed reindeer, and stick man
I was in a queue waiting to enter a party in the forest, when these ladies offered me to contribute something to the drawing they were doing together (I added stickman, but the hat and suitcase were their idea)
The most useful skill is the ability to collapse possibility into reality.
The ability to discern which possibilities you should be collapsing is what makes that skill useful
The ability to discern which possibilities you should be collapsing is what makes that skill useful
Moving to Berlin was my best decision of 2024
After a few years of knowing each other, I'm glad to announce that ChatGPT and I have upgraded our relationship to Pro tier.
Going to daytime party in teufelsberg forest old satellite station this Saturday afternoon.
Farcasters welcome to join. DM me to meet up
https://ra.co/events/2153033
Farcasters welcome to join. DM me to meet up
https://ra.co/events/2153033
Im edging towards shipping a hardware device at some point.
An agentic OS with Linux / Docker and typescript embedded, along with curated local and remote models.
Might just prototype this as a bootable USB stick.
What should I think about?
An agentic OS with Linux / Docker and typescript embedded, along with curated local and remote models.
Might just prototype this as a bootable USB stick.
What should I think about?
Payday, just banked my Farcaster UBI
Yesterday, I had the opportunity to do a pair programming session with good old integrated environment, no AI, no autocomplete.
Glad to announce I still remember how to write code without hitting tab.
Glad to announce I still remember how to write code without hitting tab.
Based thread on preserving your humanity whilst merging with AI
4790
lucas
@elesel.eth·15:54 26/04/2025
AI just told me to avoid becoming dependent on it and to stay feral (screenshots in post)
Here's how it suggests optimising for individual sovereignty over the coming decades, in 7 points, a thread:
Here's how it suggests optimising for individual sovereignty over the coming decades, in 7 points, a thread:
Been busy; want to finally ship FunWithGerman to iOS App Store by Monday.
The core feature already works and shares content to social media
Is there an experienced Swift dev hear that can help me get the app over the line and submitted to the App Store?
Will pay a 1 000 USDC bounty if you have a free weekend ahead and can get it done by Monday
The core feature already works and shares content to social media
Is there an experienced Swift dev hear that can help me get the app over the line and submitted to the App Store?
Will pay a 1 000 USDC bounty if you have a free weekend ahead and can get it done by Monday
Based Meetup. See what’s actually up in AI
533
shoni.eth
@alexpaden·17:49 25/04/2025
Weekly Farcaster AI Research Group Call:
open to the public every Tuesday-- no expertise required.
1–2 PM EDT → 11 AM–12 PM MDT, 10–11 AM PDT, 6–7 PM BST, 7–8 PM CEST, 1–2 AM CST/SGT (next day), 2–3 AM JST (next day)
Link to join: meet.google.com/aue-jdvu-rgb
https://events.xyz/f7c740
open to the public every Tuesday-- no expertise required.
1–2 PM EDT → 11 AM–12 PM MDT, 10–11 AM PDT, 6–7 PM BST, 7–8 PM CEST, 1–2 AM CST/SGT (next day), 2–3 AM JST (next day)
Link to join: meet.google.com/aue-jdvu-rgb
https://events.xyz/f7c740
Image gen now available on openAI api.
Looking at approx cost of 16 cents for a high quality 1024*1024 portrait image
https://platform.openai.com/docs/guides/image-generation?image-generation-model=gpt-image-1
Looking at approx cost of 16 cents for a high quality 1024*1024 portrait image
https://platform.openai.com/docs/guides/image-generation?image-generation-model=gpt-image-1
Lil guide from OpenAI about writing agents
https://cdn.openai.com/business-guides-and-resources/a-practical-guide-to-building-agents.pdf
https://cdn.openai.com/business-guides-and-resources/a-practical-guide-to-building-agents.pdf
Alarm clock on
828888
Scout Game
@scoutgamexyz·11:03 21/04/2025
⌛
Nice walk in Spandau Zitadel today
Since weather is getting better again, who’s up for a Berliner farcasters walk / hike next Sunday ?
Since weather is getting better again, who’s up for a Berliner farcasters walk / hike next Sunday ?
Undervalued benefit of ChatGPT compared to humans conversations.
You can always go back a step and not say the thing you just said
You can always go back a step and not say the thing you just said
Following up on yesterday's repo for learning AI agents, I just published a Substack article with a bit of walkthrough of what it's actually doing.
This should provide better context.
https://melboudi.substack.com/p/ai-agents-from-0-to-1
This should provide better context.
https://melboudi.substack.com/p/ai-agents-from-0-to-1
4339
Mo
@meb·14:21 16/04/2025
Tired of all the AI noise? I recently shipped an open source demo that takes you through the steps of agentic systems, from zero to hero.
Feedback and PRs welcome! The goal of this repo is to educate technical and non technical people alike on the components of agentic systems
https://github.com/motechFR/ai-agents-demo
Feedback and PRs welcome! The goal of this repo is to educate technical and non technical people alike on the components of agentic systems
https://github.com/motechFR/ai-agents-demo
AI Agents, from 0 to 1
There's alot of noise about AI Agents. Let's understand them from first principles.
melboudi.substack.com
Tired of all the AI noise? I recently shipped an open source demo that takes you through the steps of agentic systems, from zero to hero.
Feedback and PRs welcome! The goal of this repo is to educate technical and non technical people alike on the components of agentic systems
https://github.com/motechFR/ai-agents-demo
Feedback and PRs welcome! The goal of this repo is to educate technical and non technical people alike on the components of agentic systems
https://github.com/motechFR/ai-agents-demo
Based dev tip of the day: Ask Cursor to fix a bug by deleting code instead of adding even more code
Finally got my Apple Dev account approved this morning!
Coming soon, my first iOS app, which of course has an AI running underneath.
Will share the link.
Coming soon, my first iOS app, which of course has an AI running underneath.
Will share the link.
This is a real White House post.
Memetic branch activation of the simulation confirmed
https://x.com/whitehouse/status/1910072885789573505?s=46&t=heYf7m0kaq62UF_MFK0e6g
Memetic branch activation of the simulation confirmed
https://x.com/whitehouse/status/1910072885789573505?s=46&t=heYf7m0kaq62UF_MFK0e6g
Recent Google paper (February 2025) on prompt engineering. Super helpful
https://www.kaggle.com/whitepaper-prompt-engineering
https://www.kaggle.com/whitepaper-prompt-engineering
Nearly done with v1 of my first iOS app using Cursor + AI, coming from zero native experience as a seasoned software engineer.
What stands out:
• UI/screens are fast to build. Cursor handles layout + SwiftUI boilerplate well.
• Post-setup, bugfixing takes a few AI loops. Looking at the reasoning and debug tokens is uncanny valley territory.
• Asking the AI “why” helps me level up, but I still need docs + web deep dives.
• I now reach for docs more than before, faster than waiting on the AI sometimes.
• Prior software/product experience helps frame prompts + architecture.
• Server-side in a language I know = security peace of mind.
TL;DR:
• Pro engineers: still extremely relevant. AI boosts productivity, especially in custom, less-trodden paths.
• Copy/paste devs: rugged
What stands out:
• UI/screens are fast to build. Cursor handles layout + SwiftUI boilerplate well.
• Post-setup, bugfixing takes a few AI loops. Looking at the reasoning and debug tokens is uncanny valley territory.
• Asking the AI “why” helps me level up, but I still need docs + web deep dives.
• I now reach for docs more than before, faster than waiting on the AI sometimes.
• Prior software/product experience helps frame prompts + architecture.
• Server-side in a language I know = security peace of mind.
TL;DR:
• Pro engineers: still extremely relevant. AI boosts productivity, especially in custom, less-trodden paths.
• Copy/paste devs: rugged
Postgres really can do everything meme lives on.
Working on an app that involves generating audio. Nearly opted for classic upload / download with S3.
Then I remembered my average filesize is 50kb.
Enter postgres. I was able to create a table with a bytes column, that works really well.
PSA to keep your infra as simple as needed, and not reach out for more difficult solutions till necessary.
Working on an app that involves generating audio. Nearly opted for classic upload / download with S3.
Then I remembered my average filesize is 50kb.
Enter postgres. I was able to create a table with a bytes column, that works really well.
PSA to keep your infra as simple as needed, and not reach out for more difficult solutions till necessary.
Until recently, gpt 4o mini has been my go to for lots of agents given its low cost and latency.
Did some tests today with Gemini 2.0 flash, which somehow is even less expensive, but supposed to be full 4o parity and got amazing results, especially with it following instructions very well.
Shoutout to @alexpaden for sharing the tip on xml-ish tags for your prompts.
Seems to be more machine readable, but also makes it way easier to structure
Did some tests today with Gemini 2.0 flash, which somehow is even less expensive, but supposed to be full 4o parity and got amazing results, especially with it following instructions very well.
Shoutout to @alexpaden for sharing the tip on xml-ish tags for your prompts.
Seems to be more machine readable, but also makes it way easier to structure
Today, for the first time in a while, I solved an issue by going from stackoverflow, to Reddit, to the official docs.
Nostalgia vibes
Nostalgia vibes
First time vibe coding yesterday (swift in cursor).
Very cool experience and moved ahead incredibly fast compared to the old days of reading the docs, and moving up the stack from "hello world" when working with new languages.
I don't think all devs are cooked yet, but this has already passed the level of the average "copy paste stack overflow without caring" dev.
What I'm enjoying:
1. Being able to spend more time in product land, versus figuring out syntax.
2. Reducing the barrier to porting my engineering skills to a different language. I think LLMs will be an awesome supporter for the mobility of experienced engineering talent between language ecosystems.
3. Setting up boilerplate and screens way faster, so I can focus on iterations
Speedbumps:
1. Architecture. Getting the AI to follow rules such as decomposing screens into different files, and explain certain design decisions it makes.
2. Verbosity. Feels like too much written some times. I generally have to prompt for simplicity.
Very cool experience and moved ahead incredibly fast compared to the old days of reading the docs, and moving up the stack from "hello world" when working with new languages.
I don't think all devs are cooked yet, but this has already passed the level of the average "copy paste stack overflow without caring" dev.
What I'm enjoying:
1. Being able to spend more time in product land, versus figuring out syntax.
2. Reducing the barrier to porting my engineering skills to a different language. I think LLMs will be an awesome supporter for the mobility of experienced engineering talent between language ecosystems.
3. Setting up boilerplate and screens way faster, so I can focus on iterations
Speedbumps:
1. Architecture. Getting the AI to follow rules such as decomposing screens into different files, and explain certain design decisions it makes.
2. Verbosity. Feels like too much written some times. I generally have to prompt for simplicity.
Looking to get into iOS app development
Go for swift or react native?
Not intending to publish to android (for now)
Go for swift or react native?
Not intending to publish to android (for now)
Saved some time using AI to translate an English doc to French.
The canonical google French version for Farcaster is "Farcasteur"
The canonical google French version for Farcaster is "Farcasteur"
Feature request inspired by situation I just had;
Can we get LinkedIn post embeds like we have with Twitter?
Pros:
- Enable people targeting clients outside farcaster world to reshare content here
- Extend the farcaster social graph back into places like LinkedIn
- Build the bridge between normies and crypto natives
Can we get LinkedIn post embeds like we have with Twitter?
Pros:
- Enable people targeting clients outside farcaster world to reshare content here
- Extend the farcaster social graph back into places like LinkedIn
- Build the bridge between normies and crypto natives
