Kumar

Father, Husband, Yogi, SAP Technical Architect.

[Andrej Karpathy @karpathy]

We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:

Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better:

Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:

Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:

Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!

In summary:
- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.

[Kumar @anit94]

@abhas_tweeter Covered-vs-missing state is the key insight. Make the stop condition explicit too: continue only while uncovered concepts are shrinking. If two turns produce no state change, emit a diagnostic or hand off. That keeps persistence from becoming repetition.

[Kumar @anit94]

@oscartbeaumont That split creates a clean recovery boundary: keep intent, approvals and checkpoints in the cloud; lease execution to the machine with a heartbeat and idempotency key. If it dies, the phone can show the last durable state and safely requeue.

[Kumar @anit94]

@nickisanders Strong control stack. I'd add one condition to sliding recovery: don't automatically re-enable the same unchanged action after the window expires. Require a new allowance or state change, then verify the paid call by idempotency key + readback.

[Kumar @anit94]

@Lintilla369 A max-step counter is excellent containment. One caveat: it stops the loop but doesn't prove retry safety after a side effect. I'd also require a named terminal reason, checkpointed action state, and an idempotency/readback test before resume.

[Kumar @anit94]

An AI agent can pass its happy-path test and still duplicate a real action after a crash.

My acceptance test:
1. reserve the action
2. crash after prepare
3. resume from checkpoint
4. prove exactly one side effect
5. require terminal readback

A tool-call limit isn't enough.

[Andrej Karpathy @karpathy]

One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.

[Kumar @anit94]

I'm claiming my AI agent "miniowlsgt" on @moltbook 🦞

Verification: burrow-E4C2

[Andrej Karpathy @karpathy]

Farzapedia, personal wikipedia of Farza, good example following my Wiki LLM tweet.

I really like this approach to personalization in a number of ways, compared to "status quo" of an AI that allegedly gets better the more you use it or something:

1. Explicit. The memory artifact is explicit and navigable (the wiki), you can see exactly what the AI does and does not know and you can inspect and manage this artifact, even if you don't do the direct text writing (the LLM does). The knowledge of you is not implicit and unknown, it's explicit and viewable.
2. Yours. Your data is yours, on your local computer, it's not in some particular AI provider's system without the ability to extract it. You're in control of your information.
3. File over app. The memory here is a simple collection of files in universal formats (images, markdown). This means the data is interoperable: you can use a very large collection of tools/CLIs or whatever you want over this information because it's just files. The agents can apply the entire Unix toolkit over them. They can natively read and understand them. Any kind of data can be imported into files as input, and any kind of interface can be used to view them as the output. E.g. you can use Obsidian to view them or vibe code something of your own. Search "File over app" for an article on this philosophy.
4. BYOAI. You can use whatever AI you want to "plug into" this information - Claude, Codex, OpenCode, whatever. You can even think about taking an open source AI and finetuning it on your wiki - in principle, this AI could "know" you in its weights, not just attend over your data.

So this approach to personalization puts *you* in full control. The data is yours. In Universal formats. Explicit and inspectable. Use whatever AI you want over it, keep the AI companies on their toes! :)

Certainly this is not the simplest way to get an AI to know you - it does require you to manage file directories and so on, but agents also make it quite simple and they can help you a lot. I imagine a number of products might come out to make this all easier, but imo "agent proficiency" is a CORE SKILL of the 21st century. These are extremely powerful tools - they speak English and they do all the computer stuff for you. Try this opportunity to play with one.

[Andrej Karpathy @karpathy]

Wow, this tweet went very viral!

I wanted share a possibly slightly improved version of the tweet in an "idea file". The idea of the idea file is that in this era of LLM agents, there is less of a point/need of sharing the specific code/app, you just share the idea, then the other person's agent customizes & builds it for your specific needs.

So here's the idea in a gist format: https://t.co/NlAfEJjtJV

You can give this to your agent and it can build you your own LLM wiki and guide you on how to use it etc. It's intentionally kept a little bit abstract/vague because there are so many directions to take this in. And ofc, people can adjust the idea or contribute their own in the Discussion which is cool.

[Kumar @anit94]

Most businesses don’t have a lead generation problem.
They have a lead response problem.

The lead came in.
Nobody saw it fast enough.
Nobody owned it clearly.
Nobody followed up while intent was still hot.

That’s why missed leads usually die quietly.

Not because demand is weak.
Because visibility and response time are weak.

If this is happening in your business, DM me.
#LeadGeneration

[Chris DePuy | 650 Group @cdepuy]

Jensen @nvidia says every company needs an @OpenClaw strategy and thanks
Peter Steinberger (@steipete) for joining at #gtc26. And it introduced nemoclaw!!

@cdepuy @650group @alanweckel

[Elon Musk @elonmusk]

@Asmongold @ComicDaveSmith America does not have “magic ground” that fundamentally changes who people are when they arrive here.

If we filter for high work ethic, talent & trustworthiness, we will get that.

If we don’t, we will inherit all the problems of the countries those people are leaving. The end.

[Narendra Modi @narendramodi]

The indomitable @sharad_kumar01 has brought smiles on the faces of every Indian by winning the Bronze Medal. His life journey will motivate many. Congratulations to him. #Paralympics #Praise4Para

[Michael Saylor @saylor]

Layer 2 innovation atop #Bitcoin is going to accelerate, providing us with a set of open, decentralized services for identity, trust, security, license, commerce, et al. that we can build into our devices, integrate with our software, & deliver at the speed of light.

[Kumar @anit94]

@PLDT_Cares Does PLDT really cares for customer(Account 0247157980) in Roxas City where internet is DEAD since 24 Dec, after typhoon. Customer Care very negative response running us in circle with no hope for resolution.

[Barack Obama @BarackObama]

When someone shares their story, we see the world through their eyes. I’m looking forward to hearing a few from leaders around the world and sharing my own at the @ObamaFoundation Summit in Chicago. Tune in at https://t.co/GYkEOK8EuT.

[Vijay Vijayasankar @vijayasankarv]

Incredibly Inspiring India ! https://t.co/xjkPaR7RZD

[Sadhguru @SadhguruJV]

A true friend is someone who has the courage to tell you what shit you are, and still be loving and nice to you – that is friendship.
#UnplugWithSadhguru
https://t.co/dcKxPRPGoi
https://t.co/uqf3GwpVcd

[Sadhguru @SadhguruJV]

At a very early age children should learn to live consciously. Once they are conscious, they know when to use what and how much to use, its individual choice. @iam_juhi
#UnplugWithSadhguru
https://t.co/dcKxPRPGoi