TechnologyEvolution.AI

Husband, Father & Entrepreneur who has a passion for technology, AI, knowledge & understanding. Review new tech, AI agency, communications, data, energy.

[ClaudeDevs @ClaudeDevs]

You can now mod Claude Code:

- Change how it behaves
- Customize the UI
- Swap in your own features

Write one with a few lines of TypeScript, or have Claude build it for you. Mods ship inside plugins, so you install them with /plugin in the CLI or desktop app.

A few examples:

[TechnologyEvolution.AI @TimothiousAI]

Everyday I talk with people who don't understand whats happening in government, politics, technology, and especially in AI. It's ok, everyone lives thier own life and are on thier own journey. If it hadn't been for the finanical ruin of my own national business (thanks big red), I probably would be still using ChatGPT like I'm talking with a guy who can answer questions and write a well drafted email when I needed it, and think 'that was helpful, I like AI'. Life had a different path for me, which is a rocky road but still worth the fight and definetly worth living. My hope is to inspire people to understand whats happening at the edge, so I decided that I'd put together a list of my favorite people that I've been listening to over the past couple of years that really helped me understand AI and see where we are headed, and hope that those who are not paying attention will soon at least understand that we are literally in the middle of the singularity and want to be actively involved in shaping the future of the world as we know it.

The top of my list goes to @cole_medin , since he took a risk and left his 9-5 and started teaching people about how to code with AI, and understand whats "happening under the hood". Without his guidance I'd still be wondering how to use python and write a "hello world" script.

Next goes to @IndyDevDan, who works in the same type of domain as Cole but always seems to leave me believing I've discoverd whats at the end of the rainbow. Scale your compute to scale your impact. Truer words have never been spoken.

@natebjones is an incredible visionary and someone who has a breadth of knowledge so incredible, just listening to his 30 min hot takes help steer what I'm building and how to do it. Thanks for the time you take to drill down to understand even how different models behave and how to use them in specific workflows.

I could go on an on between @mattpocockuk, @NetworkChuck , @MatthewBerman, the latent space pod and so so many more, but there is one group of gentleman that deserve the top step, and that is @moonshots_pod with @PeterDiamandis and the truly inspirations moonshot mates @alexwg , @DaveBlundin and @salimismail, along with countless regulars and guests who are sifting through all the noise to get the true signal of whats driving innovation and the future of our world. Thank you all for the incredible work you all do each and every week you are alive. Your show drives my signal engine and helps direct everything I build and do.

My advice, tune in to the @moonshots_pod for every episode to understand how to run your business, where to invest, how to navigate life and create real impact in the agentic world we are entering. With people who truely care for others and want the best for humanity the future is bright.

Just wanted to send my gratitude for everything you all do.

[Paul Copplestone - e/postgres @kiwicopple]

who made this masterpiece ?

[Andrej Karpathy @karpathy]

We're starting to leave the territory where you'd test an LLM by e.g. "create an svg of pelican on a bicycle". As one idea to generalize it, I was interested what Opus 5 would do if I gave it the first paragraph of the Lord of the Rings, a 1M token budget (~$10) and asked for three js render of it. Opus went off for ~2 hours and wrote 5500 lines of code that (procedurally) rendered the story. It's kind of janky but fun. But it's a bit mindboggling that the LLM has to place and orchestrate various polygon assets in (x,y,z) coordinates and write code that animates it all, and that it even does anything at all.

I also like this kind of examples because no one in their right mind would ever spend the time to write something this custom but LLMs have all the stamina and patience in the world, so it's an example where we go from "no one would ever do this" to "sure, why not, it's ~free". There might be a lot more. But I'm excited about creating hyper custom worlds that you can imagine dropping players into, e.g. here to participate in the LoTR story as a spectator NPC, or one of the characters, or etc. Something like an ephemeral GTA of X on demand.

Last thought is that the domain of worlds/games exposes a weakness in LLMs: they can't easily audit their work because they aren't able to efficiently and natively perceive videos or play games within them. Here, Opus 5 had to very slowly and painstakingly take screenshots at different points, and it messed up a few times and created a bunch of jank. An example of raw capability (multimodal, gameplay) that I think is still quite lacking.

[TechnologyEvolution.AI @TimothiousAI]

@MatthewBerman We all knew it was coming. Protect your data everyone, the race is on... Next stop AGI

[TechnologyEvolution.AI @TimothiousAI]

I just gotta say, so glad GPT-5.5 (spud) was released. All my OpenClaws finally acting like they did when they were on Opus 4.7. Huge improvement over 5.4 in personality and long horizon tasks. This is the first time @OpenAI really hit the mark for me.

[TechnologyEvolution.AI @TimothiousAI]

@49agents It depends on the use case. I specify my context window / kv cache depending on what the workflow requirements. 64k is usually plenty

[TechnologyEvolution.AI @TimothiousAI]

When OpenClaw shut down I had two choices: wait for a replacement or build my own agent harness on Claude Code.
I spent 3 days building it. 13 phases. 45+ PRs. It's now more capable than what I had before.

Here's the architecture if you want to do the same:

[TechnologyEvolution.AI @TimothiousAI]

Total cost: under $200/month running on a single workstation with local inference (35B model for bulk work, 8B for classification, cloud API for orchestration and review).

If you were on OpenClaw and you're looking for what's next, this is it. Claude Code + Agent SDK + the Second Brain pattern from the https://t.co/tGmunNXBzu community. Everything you need is already there.
Follow me for more agentic processes pushing the limits of whats capable with today's technology. #technologyevolutionai

[TechnologyEvolution.AI @TimothiousAI]

The part nobody talks about: the system improves itself. Every harness run captures a trajectory (what worked, what failed, which model was used). A weekly pipeline analyzes batches of trajectories and generates evolved skill files.

Your agent literally gets better at its job every week without you touching the code (or whenever you want)!

[TechnologyEvolution.AI @TimothiousAI]

I built this into my AI Second Brain today.

Karpathy is right, at personal scale, a well-maintained index.md beats vector search. The LLM just reads it and knows where to look.

The part nobody's talking about: this same pattern scales. Add hybrid RAG at 500+ docs and it becomes an enterprise knowledge system.

[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.

[TechnologyEvolution.AI @TimothiousAI]

Can someone tell Anthropic it’s not Christmas… At least the gov service is green. Sheesh

[Andrej Karpathy @karpathy]

Software horror: litellm PyPI supply chain attack.

Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords.

LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm.

Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks.

Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages.

Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.

[TechnologyEvolution.AI @TimothiousAI]

@MatthewBerman 😂

[TechnologyEvolution.AI @TimothiousAI]

@andyreed ok - I have had a good laugh out loud moment in a while. Not at the actual poor bloke who got hurt, but at all the subagent experiences I've had so many times this is the case. Claude shipped agent teams (experimental) but its the right direction.

[Andrej Karpathy @karpathy]

+1 for "context engineering" over "prompt engineering".

People associate prompts with short task descriptions you'd give an LLM in your day-to-day use. When in every industrial-strength LLM app, context engineering is the delicate art and science of filling the context window with just the right information for the next step. Science because doing this right involves task descriptions and explanations, few shot examples, RAG, related (possibly multimodal) data, tools, state and history, compacting... Too little or of the wrong form and the LLM doesn't have the right context for optimal performance. Too much or too irrelevant and the LLM costs might go up and performance might come down. Doing this well is highly non-trivial. And art because of the guiding intuition around LLM psychology of people spirits.

On top of context engineering itself, an LLM app has to:
- break up problems just right into control flows
- pack the context windows just right
- dispatch calls to LLMs of the right kind and capability
- handle generation-verification UIUX flows
- a lot more - guardrails, security, evals, parallelism, prefetching, ...

So context engineering is just one small piece of an emerging thick layer of non-trivial software that coordinates individual LLM calls (and a lot more) into full LLM apps. The term "ChatGPT wrapper" is tired and really, really wrong.