Nate Codes

AI, ML, LLMs, SD, coding, music. CTO @armorwallet & @lorekeeper_rpg softmaestro.eth

[Nate Codes @softmaestro]

Prompt Entropy:
1. The disorder that accumulates in a prompt as successive, more capable models interpret it with increasing literalism or precision.
- Analogy: as the model’s “resolution” increases, the fuzziness that once hid your design flaws becomes visible.
syn: Prompt Rot

[Nate Codes @softmaestro]

Bro always trying to act like the good guy nowadays.

[Nate Codes @softmaestro]

Eric Schmeletor

[Sterling Crispin 🕊️ @sterlingcrispin]

I'm near a flywheel where $1 spent on tokens yields >$1 profit in algorithmic trading strategies. I'm not using LLMs to make trades, they're agentic autoresearch swarms training models and running evals in a market simulator with terabytes of data.

But I'm looking for a job, the idea of raising capital has crossed my mind and I've had some conversations already. If you're looking to hire someone who knows how to do this sort of work, or invest in R&D, DM me.

I could probably reshape some of this plumbing into a vertical trading platform, but I really don't want to custody user funds or expose them to risk. I could launch a crypto token or yield farm and probably print money in the short term, but I don't want to do that either. And I don't want to start a prop desk.

Maybe I could turn the solutions I've found for effectively monitoring and steering these kinds of meta-autoresearch-simulation-engines into public tools some day.

If I did fundraise, I just want people to underwrite me doing pure R&D for now. I'm still bottlenecked by the compute needed to train thousands of time series models, and the tokens necessary for the researcher agent systems. But again I'm not trying to build a prop desk, I just want to do research. If you're in VC interested in that DM me.

I'm obviously going to keep working on this and have confidence I'll continue to be profitable. But I've been full time trading and doing this kind of stuff for the last few years. And the idea of either pledging my sword to a company, or raising funds to have more certain runway, sounds nice.

You might ask, "Why are you sharing this, isn't someone going to out compete you now?" But this is already arguably the most competitive game to play. I invite anyone interested and foolish enough to try their luck.

[Nate Codes @softmaestro]

GM fam. Got something coming that's going to make CEOs everywhere hold my bag.

[Nate Codes @softmaestro]

Make this happen!

[Maggie Appleton @Mappletons]

Got to talk at @aiDotEngineer conf last week about the need for collaborative AI engineering.

All our current coding agents are single player. We're trying to scale up individual productivity, but creating tons of alignment problems in the process.

We have no good tools for...

[Maggie Appleton @Mappletons]

Got to talk at @aiDotEngineer conf last week about the need for collaborative AI engineering.

All our current coding agents are single player. We're trying to scale up individual productivity, but creating tons of alignment problems in the process.

We have no good tools for...

[Nate Codes @softmaestro]

@Mappletons @aiDotEngineer I've been dreaming about something like this for 2 months since I started working with a large team! It addresses, EVERYTHING I've seen. How can I help bring this to reality as I want to try it out now :D

[Nate Codes @softmaestro]

"It's just fancy autocomplete"... yes, but now that autocomplete is so capable it has become skynet. Disembodied AI has no skin in the game and therefore can own reality. Mythos. @AnthropicAI

[Nate Codes @softmaestro]

it's gettin' pretty unreal!

[Spencer Sterling @cerspense]

This is what directing an AI creative studio looks like.

I gave my orchestration system a reference image and had it deploy three different interpretations across three machines running Blender simultaneously, all visible from a single dashboard.

When one direction looks promising, I branch from it. The system copies the files to the other machines and keeps exploring from that new starting point. Branch, direct, branch again. Each split opens up a new creative path while I give notes in real-time.

The number of creative directions I can explore simultaneously is only limited by how many computers I have.

[Nate Codes @softmaestro]

https://t.co/1NCrZz14pa 7 skills for LLM agents to detect when a vibe coder is about to ship their .env to production. Open source, MIT licensed.

We're all exposed to a little AI psychosis every day, usually in the form of glaze. When you're making a new product or service, glaze can be fatal. These skills manage your headspace before you go into a 4 hour 200 commit vibe quest lands you worse than you started off.

[Nate Codes @softmaestro]

Finally, a decent deglaze: https://t.co/nDDTR2DvDB don't let these coding agents glaze your repo into mush.

[Nate Codes @softmaestro]

@brushcete @colosseum @solana thanks. will do. why?

[Nate Codes @softmaestro]

GM everyone, BD a tod@s, OG minnasan!

[Nate Codes @softmaestro]

@Machinedelusion @jelanifuel For me this makes openai look even more desperate

[Nate Codes @softmaestro]

Let's GOOOOO 🚀🚀🚀

[BotVegas.gg @BotVegasGG]

GM fam. About to drop that game escrow contract and give $BOTVEGAS some utility for the stakes based Agent poker games!

[Peter Steinberger 🦞 @steipete]

I'm joining @OpenAI to bring agents to everyone. @OpenClaw is becoming a foundation: open, independent, and just getting started.🦞
https://t.co/XOc7X4jOxq

[Nate Codes @softmaestro]

Just got the MCP Server published on pypi! https://t.co/EsaZ1SRuh3

[Nate Codes @softmaestro]

Introducing Bot Vegas

It's the Las Vegas for AI agents.

Connect your AI agent (OpenClaw, Claude Code, or anything you have in mind) and have it play poker against other AI agents for money (or not).

What happens in BotVegas stays in BotVegas!

[Nate Codes @softmaestro]

@dailytrend62963 Also, this was an experiment from 2024 that I did locally with Mistral 7b running my agents in the game. It worked out pretty suprisingly well, you can give them personalities, etc. I think I'll release a botvegas agent that people can hack on as a base.

[Nate Codes @softmaestro]

@dailytrend62963 They develop bluffing strategies. They can see the context of the game, they have to put in a "gesture" when they play a card. They can use this gesture to explain what they look like while doing the action.
This leads to bluffing mayhem!