wali.eth

Software Engineer

[Julián @juliandeangeIis]

Siendo que estoy todo el tiempo probando tools de AI, solo tuve pocos momentos 'wow':
- Cursor (Feb 25)
- Claude Code (Ag 25)
- Cloud Agents + SDD + Automations (Sep-Oct 25)

y sin dudas de las últimas semanas tengo uno nuevo:
- Hermes + Obsidian + LLM Wiki (skill de karpathy)

[Alex Song @alexsongis]

A lot of guys hearing about Jane Street for the first time. Lol.

Trust me when I say Jane Street does not have a brand awareness problem.

It has 100% awareness penetration in its talent pool because of this:

iykyk

[Brandon Luu, MD @BrandonLuuMD]

Literally just having a delusional golden retriever mindset measurably changes outcomes and physiology.

Sleep badly? Convince yourself you're well rested.
Stressful day? Convince yourself it's fuel.
Failed? Convince yourself it's useful data.

[Oliver Prompts @oliviscusAI]

Someone built a Claude Code skill that strips every trace of AI from your writing.

It detects 29 specific patterns from Wikipedia's "Signs of AI writing" guide, em dashes, rule of three, "it's not just X, it's Y", sycophantic openers, and rewrites your text to actually sound human.

100% Open Source.

[How To Prompt @HowToPrompt__]

MIT proved every major AI model is secretly converging on the same "brain."

It’s called the “platonic representation hypothesis,” and it’s one of the most mind-blowing papers you’ll ever read.

You train a vision model purely on images. You train a language model purely on text.

They use completely different architectures. They process completely different data. They should have completely different "brains."

But as these models scale up, something impossible is happening.

When researchers measure how they organize information, the mathematical geometry is identical.

A model that only "sees" images and a model that only "reads" text are measuring the distance between concepts in the exact same way.

The models are converging.

The researchers named this after Plato’s Allegory of the Cave.

Plato believed that everything we experience is just a shadow of a deeper, hidden, perfect reality.

The paper argues that AI models are doing the exact same thing.

They are looking at the different "shadows" of human data, text, images, audio. And they are independently discovering the exact same underlying structure of the universe to make sense of it.

It doesn't matter what company built the AI.

It doesn't matter what data it was trained on.

As models get larger, they stop memorizing their specific tasks. They are forced to build a statistical model of reality itself.

And there is only one reality to map.

2024, Arxiv

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[elvis @omarsar0]

NEW AI report from Google.

Every prior intelligence explosion in human history was social, not individual.

These authors make the case that the AI "singularity" framed as a single superintelligent mind bootstrapping to godlike intelligence is fundamentally wrong.

This is directly relevant to anyone designing multi-agent systems.

They observe that frontier reasoning models like DeepSeek-R1 spontaneously develop internal "societies of thought," multi-agent debates among cognitive perspectives, through RL alone.

The path forward is human-AI configurations and agent institutions, not bigger monolithic oracles.

This reframes AI scaling strategy from "build bigger models" to "compose richer social systems."

It argues governance of AI agents should follow institutional design principles, checks and balances, role protocols, rather than individual alignment.

Paper: https://t.co/bfwrnbkY2y

Learn to build effective AI agents in our academy: https://t.co/1e8RZKs4uX

[Jaynit @jaynitx]

In 2019, MIT professor Patrick Winston gave a legendary 1-hour lecture called “How to Speak.”

It has 18M+ views for a reason.

His frameworks:
• Your ideas are like your children
• The 5-minute rule for job talks
• Why jokes fail at the start

15 lessons on communication:

[Ryan Carniato @RyanCarniato]

I never thought this day would come.

Thanks to AI, we've hit the inversion point where TDD is something that actually saves time instead of wastes time.

What a world we live in.

[mrinank @MrinankSharma]

Today is my last day at Anthropic. I resigned.

Here is the letter I shared with my colleagues, explaining my decision.

[Argona @Argona0x]

i gave an AI $50 and told it "pay for yourself or you die"

48 hours later it turned $50 into $2,980

and it's still alive

autonomous trading agent on polymarket

every 10 minutes it:

→ scans 500-1000 markets
→ builds fair value estimate with claude
→ finds mispricing > 8%
→ calculates position size (kelly criterion, max 6% bankroll)
→ executes
→ pays its own API bill from profits

if balance hits $0, the agent dies

so it learned to survive

built in rust for speed

claude API for reasoning (agent pays for its own inference)

runs on a $4.5/month VPS

weather markets: parses NOAA before polymarket updates sports: scrapes injury reports, finds mispricing crypto: on-chain metrics + sentiment

$50 → $2,980 in 48 hours

how much do u think i’ll see in a week?

[Simplifying AI @simplifyinAI]

Goldman Sachs is rolling out Claude to automate accounting completely.

After 6 months of embedded engineering, they found that AI reasoning capabilities aren't limited to coding.

The same "step-by-step" logic that fixes Python bugs works perfectly for:

→ Complex accounting rules
→ Trade reconciliation
→ Compliance exceptions

They are now deploying "digital co-workers" to handle client onboarding and vetting faster than humanly possible.

Goldman says this will slow headcount growth rather than cause immediate layoffs.

[Simplifying AI @simplifyinAI]

Docker for AI Agents is officially over..

Pydantic just dropped Monty. It's a python interpreter written in rust that lets agents run code safely in microseconds.

no containers. no sandboxes. no latency.

100% open source.

[Astrid Wilde 🌞 @astridwilde1]

all of you owe Yann LeCun an apology. the man is about to be directly responsible for saving tens of thousands of future lives

[DCinvestor @DCinvestor]

vibe coders should understand something:

i love how easy AI is making it for people to build their own apps, push them into production, and start businesses

but let's be clear: the future is not in humans building consumer-facing apps

the future is everything becomes an API which your personal AI agent can interact with in ways which suit your specific needs and lifestyle (down to the very specific needs of you as an individual)

the fact that you can use the machines to build your apps is just an intermediate step to the machines creating the apps for you, LIVE, as you need them

so the value of you learning how to build apps now really lies in you learning how to create a business model behind that app- not in creating the piece of software that is the app itself

sure, there will be templates for how you can interact with those apps/APIs, but your personal AI will pick one and tailor it even further for you. and a lot of the time, you won't even need to interact with a UI beyond speaking with your AI assistant

let me give you an example: would you rather use an app like Uber or Uber Eats, or would you rather just ask your AI assistant to get you a ride somewhere or to show you menus for the type of food you might be interested in and you pick one? the value in apps like that is not in the app installed on your phone. it's in the backend business model which connects the customer with providers. and personal AI assistants actually open the door to you being able to seamlessly use multiple business APIs without worrying in the slightest about which app or intermediate provider they come from

there is a decent chance apps as you know them will be mostly dead in ~5-10 years

and yes, there are some apps which will still require deep optimization and that is where the hardcore coders may still be needed. but machines will get better at that, and if you take one look at the AAA gaming landscape, you should understand that hyper-optimized code isn't as valuable as it used to be

but what will be valuable is owning the APIs with the most use and liquidity. and yes, a lot of those will use public blockchains

things are going to accelerate and get very weird very quickly from here

[Claude @claudeai]

Ads are coming to AI. But not to Claude. Keep thinking.

[Yann LeCun @ylecun]

@elonmusk @farzyness Actually, quite the opposite.
I know I can do it and I know how to do it.
Just not with the techniques everyone is currently betting on.
My bet is (famously) on JEPA, world models, and planning.
At some point, you'll realize I'm right 😅

[Andrej Karpathy @karpathy]

A few random notes from claude coding quite a bit last few weeks.

Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words. It hurts the ego a bit but the power to operate over software in large "code actions" is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I'd expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent.

IDEs/agent swarms/fallability. Both the "no need for IDE anymore" hype and the "agent swarm" hype is imo too much for right now. The models definitely still make mistakes and if you have any code you actually care about I would watch them like a hawk, in a nice large IDE on the side. The mistakes have changed a lot - they are not simple syntax errors anymore, they are subtle conceptual errors that a slightly sloppy, hasty junior dev might do. The most common category is that the models make wrong assumptions on your behalf and just run along with them without checking. They also don't manage their confusion, they don't seek clarifications, they don't surface inconsistencies, they don't present tradeoffs, they don't push back when they should, and they are still a little too sycophantic. Things get better in plan mode, but there is some need for a lightweight inline plan mode. They also really like to overcomplicate code and APIs, they bloat abstractions, they don't clean up dead code after themselves, etc. They will implement an inefficient, bloated, brittle construction over 1000 lines of code and it's up to you to be like "umm couldn't you just do this instead?" and they will be like "of course!" and immediately cut it down to 100 lines. They still sometimes change/remove comments and code they don't like or don't sufficiently understand as side effects, even if it is orthogonal to the task at hand. All of this happens despite a few simple attempts to fix it via instructions in CLAUDE . md. Despite all these issues, it is still a net huge improvement and it's very difficult to imagine going back to manual coding. TLDR everyone has their developing flow, my current is a small few CC sessions on the left in ghostty windows/tabs and an IDE on the right for viewing the code + manual edits.

Tenacity. It's so interesting to watch an agent relentlessly work at something. They never get tired, they never get demoralized, they just keep going and trying things where a person would have given up long ago to fight another day. It's a "feel the AGI" moment to watch it struggle with something for a long time just to come out victorious 30 minutes later. You realize that stamina is a core bottleneck to work and that with LLMs in hand it has been dramatically increased.

Speedups. It's not clear how to measure the "speedup" of LLM assistance. Certainly I feel net way faster at what I was going to do, but the main effect is that I do a lot more than I was going to do because 1) I can code up all kinds of things that just wouldn't have been worth coding before and 2) I can approach code that I couldn't work on before because of knowledge/skill issue. So certainly it's speedup, but it's possibly a lot more an expansion.

Leverage. LLMs are exceptionally good at looping until they meet specific goals and this is where most of the "feel the AGI" magic is to be found. Don't tell it what to do, give it success criteria and watch it go. Get it to write tests first and then pass them. Put it in the loop with a browser MCP. Write the naive algorithm that is very likely correct first, then ask it to optimize it while preserving correctness. Change your approach from imperative to declarative to get the agents looping longer and gain leverage.

Fun. I didn't anticipate that with agents programming feels *more* fun because a lot of the fill in the blanks drudgery is removed and what remains is the creative part. I also feel less blocked/stuck (which is not fun) and I experience a lot more courage because there's almost always a way to work hand in hand with it to make some positive progress. I have seen the opposite sentiment from other people too; LLM coding will split up engineers based on those who primarily liked coding and those who primarily liked building.

Atrophy. I've already noticed that I am slowly starting to atrophy my ability to write code manually. Generation (writing code) and discrimination (reading code) are different capabilities in the brain. Largely due to all the little mostly syntactic details involved in programming, you can review code just fine even if you struggle to write it.

Slopacolypse. I am bracing for 2026 as the year of the slopacolypse across all of github, substack, arxiv, X/instagram, and generally all digital media. We're also going to see a lot more AI hype productivity theater (is that even possible?), on the side of actual, real improvements.

Questions. A few of the questions on my mind:
- What happens to the "10X engineer" - the ratio of productivity between the mean and the max engineer? It's quite possible that this grows *a lot*.
- Armed with LLMs, do generalists increasingly outperform specialists? LLMs are a lot better at fill in the blanks (the micro) than grand strategy (the macro).
- What does LLM coding feel like in the future? Is it like playing StarCraft? Playing Factorio? Playing music?
- How much of society is bottlenecked by digital knowledge work?

TLDR Where does this leave us? LLM agent capabilities (Claude & Codex especially) have crossed some kind of threshold of coherence around December 2025 and caused a phase shift in software engineering and closely related. The intelligence part suddenly feels quite a bit ahead of all the rest of it - integrations (tools, knowledge), the necessity for new organizational workflows, processes, diffusion more generally. 2026 is going to be a high energy year as the industry metabolizes the new capability.

[ℏεsam @Hesamation]

this is banger. if you use Claude Code, start using Flowy for the planning phase. it replaces Claude’s ascii chart with a UI graph you can customize and send back to AI to show exactly what you want.

AI coding will be transformed by simple creative techniques like this.

[Alec Helbling @alec_helbling]

Functions are Vectors

This is a great blog post from Max Slater about how viewing functions as infinite dimensional vectors unlocks powerful tools from linear algebra.

I think this is an under-emphasized perspective in intro linear algebra classes.