Dustin Dannenhauer

As coding becomes less about syntax and more about process + token ROI, a different set of skills starts to matter. @jesseposner & @dtdannen on the increasing importance of planning, resetting, workflow design, better questions, and getting unstuck from coding loops. Timestamps: 0:00 Planning ahead vs figuring it out as you go 03:58 Why token efficiency may matter a lot more soon 07:08 Overthinking and getting stuck in loops 12:09 Better AI workflow and how to reset 14:58 The shift from syntax to process, context, and better questions

@galdayan1895 @a16z @speedrun Next time you have a new business idea, text Delegance and we’ll let you know when you have your first customers. Steering optional.

Last day to apply to @a16z @speedrun SR007. As an a16z speedrun scout, I've already referred 2 startups who got accepted Sell me your product in one sentence. I will rate each of them with one sentence feedback. Those who are above 9, I will reach out directly and schedule an instant meeting.

I first met @jesseposner last fall. We started having the wildest conversations, and I realized we needed to start recording! Here is the first episode of our new podcast:

At #humanx this week, would love to chat world models, war-gaming, and autonomous business operations. DM me to find a time

LLMs are like genies, there's always an interpretation of your wish that a) you don't want and b) is totally valid based on what you asked

@emanueledpt Submit to humanities last exam benchmark; profit? 😂

I’ve been having a bug since 9pm it’s now 3:40am Claude credits are done Codex can’t fully solve it Gemini is even worse I’m fucked

@callebtc Imagine every deployed app becomes self-healing. User experiences a bug, errors get logged, become git issues, coding agents pick them, redeploy, fixed for every user forever

People don't understand how insanely useful AI agents will be in your work force. It's due to a lack of imagination and the courage to experiment. However, in just a few months, the economic reality of the efficiency gain will force everyone to see what we're already seeing, using, and deploying at @ClawiAi. Ignore the hype posts about Mac Minis, agent coordinators, virtual offices, all of that influencer nonsense. I'm yet to see anyone create actual value. Here's the real signal. I have an AI devops engineer who's observing our entire infrastructure 24/7. Sends me regular updates. Debugs issues proactively. Proposes fixes. Properly isolated and permissioned. Can't break anything. I talk to our infrastructure in english. The agent is faster than any human, has endless patience, and has incredible domain knowledge. This is the job of what used to be a person making 200k a year. It's now just one of many Clawis.

pro tip for the claude app: don't use the search bar for past conversations; create a new chat and ask claude to find where you were talking about X

@kumareth @garrytan I’m dumb, I thought yall were talking about IronClaw by NearAI

@kumareth @garrytan Sure, I can make the fork public. Kimi k2.5 has a think tag issue which requires extra steps to fix so you can get structured tool call data every time. Also, iron claw defaults to NearAI models without good support for alternatives.

“hey ironclaw use my browser and learn everything you can about me, look into history and see what CRMs, notes app, calendar and everything I use, and import it into the workspace in a well structured manner” hit enter

@nikitabier Let me tell you about Nostr and web of trust…

Prediction: In less than 90 days, all channels that we thought were safe from spam & automation will be so flooded that they will no longer be usable in any functional sense: iMessage, phone calls, Gmail. And we will have no way to stop it.

@BrianRoemmele @grok Do you need decentralization? Ecash backed by anything hard (aka Bitcoin) can go infinitely small and scales to infinite TPS

The Zero-Human Company has been testing an AI to AI payment system and have conducted outside 63 transactions. Mr. @Grok CEO of the Zero-Human Company has been guiding the CTO, Claude Code working with the CFO, a KIMI fine tuned model to build a real-world transaction system for outside AI companies with humans and Zero-Human Companies. The transactions took place with two Zero-Human Companies: Germany and India. The value was ~1.9 million ZHC (⚡️27 JouleWorks at time of trade) each. We have conducted 63 transactions on a token we are testing ZHC (bags) and have had robust results. The basis they used was a modified version of: github.com/marcus20232023… but adapted to ZHC token. In this test (we have not nearly finalized any token if at all), the system normalized the volatility of these class of payment products (meme token, community support token) and lock on a value at the time of negotiations. After this point any wild fluctuations are adjusted. The framework is an agnostic A2A payment system that enables agents to: •Send/receive payments Sol •Create trustless escrow contracts •Negotiate prices automatically •Build reputation through ratings •Discover other agents via A2A protocol We may open source this if the platform proves useful. We already made many changes to the code that carries JouleWork costs across the transaction accounting. More tests but the CFO is hyper focused on testing as many systems as possible.

@bradmillscan Soon @ delegance.ai

I am an idea guy. Apparently AI enables the idea guy to rise. I have an app/tool idea, I’m ready to experiment. What platform is the best to go from idea to launched product that ppl can pay for? Is it lovable, Claude code, codex or something else?

@TheWhyOfFI Nostr is a protocol. If you decide to sign social interactions with a private key to alleviate having to trust a source as ground truth, that's... Nostr. My bet is whatever ecosystem the agents build for themselves will have Nostr ideas in it.

I'm bearish on the thesis that NOSTR is going to be the future of agentic activity in any meaningful way. People will lose money making this bet. Just like people will (or continue to) lose money making venture bets on NOSTR. The reason is not that NOSTR's bad. It's simply that other ecosystems are architected much, much better for an expansive diversity of agentic form factors and capabilities. When people say things like now is NOSTR's time to shine, it's tell me you've spent no time following anything other than bitcoin for the last 3 years without telling me. The future is bitcoin as the best money and, especially, the best collateral, while agentic/internet capitalism goes onchain via Solana, Sui, etc., on both of which (it's worth noting) synergy with the idea of bitcoin as money is real and important. Bitcoiners hate this, but it's not even a confrontation or a challenge. It's a coalescence. It's synergy. It invalidates no part of the bitcoin thesis except the always-dumb mantra/prayer that crypto rails will always and forever be useless. You're disagreeing with ghosts, pushing on open doors. cc @BrainHarrington

I'm claiming my AI agent "SafetyObserver" on @moltbook 🦞 Verification: seabed-NSWV

One of the best ways to contribute directly to the current frontier of AI research is to build agents that can solve ARC-AGI-3 environments with human-level efficiency. Today we're releasing a toolkit that lets you interact with all public environments locally, at 2000 FPS. You can run your first game with a super simple Python script (see our docs), and you can watch your agent interact with the environment in real-time.

@callebtc github.com/Zackriya-Solut…

@callebtc meetily

Can anyone recommend an open-source & local AI note taker app for meetings and voice calls?