Jungmin Park @arenaaaa

Full Jev Tutorial What it is, how you can build with it and what new applications it can unlock → 0:00 Intro → 0:34 Jev explained → 4:06 API setup → 5:59 Demo 1: Voice-controlled browser → 11:33 Demo 2: AI memory → 17:27 Demo 3: YouTube predictor

This is f*cking gold.. Chinese researchers just published a paper with a brutal title: "The End of Prompt Engineering". The argument is simple: telling agents what to do is finished as a skill. A prompt is a wish, and a wish repeated three hundred times is still a wish. Boundaries written in a file are law -- read by every agent before every action, versioned, owned by you. At swarm scale nothing else composes. Three hundred agents cannot share a vibe, they can share a file. So the person stops being the one writing instructions. You turn into the boundary architect -- you define what never happens and audit the asks. Prompts turn disposable, the constraint file turns into the job. That is exactly the system I built and documented below: one AGENTS.md, three hundred Kimi K3 readers, zero escapes in 41 days. The paper's whole thesis in one line: write the fence before the keys ↓

slash1s @slash1sol

a month ago

x.com/i/article/2097…

25 7 146 554K 225

In 1986 a guy got kicked out of every casino in Vegas for counting cards. So he flew to Hong Kong with $180,000 and started betting on horses instead. He walked away with almost $900 million. It's Bill Benter. He figured horse racing was just another counting problem. Same math, more moving parts. He and a partner showed up with $180k and a computer. Benter spent years teaching that computer to guess one thing, the real chance each horse had to win. If his number was better than the odds the bookies gave, he bet. If not, he skipped it. That's the whole trick. Expected value. EV = p · b − (1 − p) Only bet when your win chance p, at odds b, is worth more than your chance of losing. This recording was never meant to be some hidden gem. Nobody expected Professor Tsitsiklis to hand the whole foundation away in 45 minutes, but that's exactly what happens on the board. Students in that room pay over $80,000 a year to sit through it. It's free right here. It's free right here. Every quant, every professional bettor, every hedge fund analyst started with this exact hour. Benter just watched it and actually did the homework. Almost nobody knows this lecture even exists. Watch it before it gets taken down. The answer is in this video.

We will give one banked reset for every day you don't have access to Astra on your paid ChatGPT plan, starting today. Team is moving mountains to give access as fast as we can. First one will land in ~ 3 hours. There is still time to create your account if you don't have one.

This is truly a must-read. Please read it, and once you have, bookmark it and revisit it often. Written by one of the smartest people I know, it offers powerful insights into the future ahead. All the more essential if you’re an investor.

fin @fi56622380

2 months ago

x.com/i/article/2092…

66 436 2K 984K 7K

🚨BREAKING: MIT researchers discover how to enable LLMs to do real logical reasoning. This is what you need to know: (thread)

Google's team just dropped a 9-page PDF on Harness Engineering - one formula replaced prompt engineering: the twist: same Claude Sonnet, same benchmark - change only the harness the formula: Agent = Model + Harness here's the playbook in 6 steps: step 1 → add guides - AGENTS.md, rule files, constraint docs - each line is a past agent failure turned into a permanent fix step 2 → add sensors - linters, tests, validation scripts the agent runs on its own output before a human sees it step 3 → build the agentic loop - plan, execute, verify, fix - bounded retries, budget caps, escalation when stuck step 4 → externalize memory - the model forgets every session. the harness remembers state, decisions, and artifacts across all of them step 5 → enforce permissions - which tools, how many writes, what needs approval. the model doesn't enforce safety. the harness does step 6 → wire observability - track every tool call, cost, retry. trip wires fire when behavior drifts the result: your agent stops being a demo and starts being infrastructure - every failure makes the system permanently better, not just the next conversation this 9-page PDF is what comes after prompt engineering and context engineering bookmark and build the full system from the article below ↓

chuplung @choopyplug1

2 months ago

x.com/i/article/2088…

26 108 564 1.3M 2K

@dwarkesh_sp 1 GW can produce 28 BTC today = $2.2 Million. The same GW costs $15B to buy and can generate $50B revenue. What's the point of BTC as a store of value when energy itself can do that much better and doesn't even have to be limited to a certain number? (21M)

This Meta and Stanford University paper is f*cking fantastic Researchers just proved that code is the operating system for autonomous agents A new research paper proves that treating code as an executable harness rather than raw text output unlocks verifiable long-horizon agent autonomy From: code as a final generation output → To: code as an executable runtime harness macro objective → programmatic planning → sandboxed execution → runtime sensor verification → multi-agent state convergence Meta and Stanford established a unified 3-layer architecture for stateful and self-evolving agent systems: → Layer 1: Executable Harness Interface Replaces free-form text reasoning with program-delegated execution, allowing interpreters and formal symbolic solvers to verify intermediate logic → Layer 2: Feedback Loops & Deterministic Sensors Enforces sandboxed permission boundaries and test sensors that convert compiler errors and execution traces into automated repair signals → Layer 3: Multi-Agent Shared Substrate Coordinates specialized agent pools (planners, coders, verifiers, and critics) over shared Git repositories and blackboard states to eliminate coordination drift Hypothesis Verification & Empirical Results Meta and Stanford hypothesized that grounding agent reasoning, state, and coordination in executable code harnesses eliminates compounding error drift in multi-step workflows. Practical validation across repository-level benchmarks confirmed the findings: code-harnessed execution drove an 84.2% task success rate (+119% gain over unmanaged baseline agents) while reducing multi-agent state desynchronization by 73% Building production AI agents is not about writing clever system prompts. The leverage comes from engineering the code harness that sandboxes execution, verifies runtime sensors, and turns raw models into deterministic systems Read the complete breakdown in the article below ↓

keyboard_arrow_left Previous keyboard_arrow_right Next

marfin @marfinxx

3 months ago

x.com/i/article/2081…

13 58 421 583K 1K

Neoclouds including SpaceX are expected to rival the hyperscalers in size in ‘27 …adding 5-10x current capacity $AMZN $MSFT $GOOGL $META vs. $SPCX $NBIS $CRWV

x.com/i/article/2089…

Useful @EpochAIResearch piece on how increasingly complex AI financing structures actually work. The Anthropic deals are a good example: >Compute: SPV buys >1 GW of TPUs → leases to Anthropic → institutional capital funds ~$35B → Broadcom backstops most senior debt >Data centers: project SPVs fund construction → Fluidstack leases capacity → Anthropic uses it → Google backstops portions of the rent The stack increasingly looks like: AI demand → long-term contract → SPV/project company → institutional capital → vendor/hyperscaler credit support Where I disagree slightly is the conclusion that financing therefore isn't a constraint. >These deals show you can finance incremental 1–2 GW projects. But we're talking about potentially mid-teens GW of incremental capacity in North America and ~30 GW globally. That quickly becomes hundreds of billions to >$1T of capital! At that scale, the marginal decision-maker influences outcomes and the controlling party is shifting. It is no longer just convincing hyperscaler C-suites that AI has enormous long-term value. You increasingly have to convince banks, insurers, bond funds and the marginal syndicated lender that these cash flows are durable enough to repay them. So, I would frame it less as “financing isn't a constraint” and more as financial engineering has pushed the constraint outward. We also don't know a world where these SPVs blow up and hyperscale/OEMs get enforced. Have a feeling these guys negotiated much more leverage than they let on while promoting these vehicles...

keyboard_arrow_left Previous keyboard_arrow_right Next

Mizuho on $AVGO / $LITE / $CRDO / $NVDA etc. Key points: 1) strong VR200 NVL72 ramps in 2027E, ramping 4Q26E led by xAI and META, as NVDA potentially delayed 4-die transition to Feynman with VR-Ultra on 2-die configuration, 2) we see focus on driving performance with NPO/CPO connectivity, ramping Spectrum-X (positive for LITE), 3) we see HBM content/GPU from GB300 to VR200 flat at 288GB while VR-ultra moves to 256GB HBM4e, but 70-100% higher HBM4e pricing and Feynman potentially returning to HBM content growth, 4) 2027E CoWoS growing >75% with NVDA/AVGO, and volume CoPoS ramp in 2028E, 5) INTC EMIB-T tailwinds from MTK's TPU being the first high-volume customer 2028E, and 6) CSP RPO/backlog now at $2.3T (up 3.5x y/y) as AgenticAI/inference accelerates. We remain positive with potential upside to DELL/CRDO/NVDA/AVGO/LITE in 2027E.

Sean @sean_________

2 months ago

Mizuho CoWoS/GPU #'s Qtrly call, 8/14: $NVDA, $DELL, $SMCI, $AVGO, $CRDO, $TSM, $AMKR

keyboard_arrow_left Previous keyboard_arrow_right Next

2 14 113 44K 105

THE MOST PROFITABLE AI SEMICONDUCTOR STOCKS Operating margin breakdown: • Best-in-class (50%+) | $MU, $NVDA, $SNDK, $SKHY, $TSM • Elite (40–49%) | $ANET, $AVGO, $KLAC • Strong (20–39%) | $ASML, $LRCX, $WDC, $CRDO, $AMAT, $ALAB • Solid (Below 20%) | $ARM, $LITE, $MRVL, $AMD, $COHR, $ON

Anthropic engineer: "90% of our engineers were already running self-improving loops Now everyone is moving toward agentic graphs" "Prompting is basically finished" In just 10 minutes, she builds her complete Claude Code setup live from an empty terminal Agents → Loops → Graphs → Self-Improving Systems Prompting was the old workflow Graph engineering is the next one This 10-minute breakdown is worth more than most $1,000 agent engineering courses Watch it today Then save the full guide below before everyone catches on ↓

rari🈀 @0xwhrrari

2 months ago

x.com/i/article/2086…

37 179 1K 4.2M 4K

Stanford just released the kind of LLM lecture frontier AI labs pay engineers $750K+ a year to understand Not prompting Not RAG How to actually build language models from scratch In 104 minutes, they break down the full stack behind modern LLMs: 06:18 - how raw data becomes a language model 21:42 - tokenization and model architecture 39:17 - why scaling laws can predict model performance 57:36 - how compute, data, and parameters trade off 1:14:28 - what really happens during post-training 1:34:11 - the engineering skills behind frontier models This one lecture is more useful than 100 random LLM tutorials Bookmark it and watch it today Then read the full guide below and learn how to build language models from scratch

Lunar @LunarResearcher

3 months ago

x.com/i/article/2086…

18 105 569 2.0M 2K

Head of Claude Code: "I'm not prompting my agents anymore, I'm building graphs and loops so they can build the agents for me." In 10 minutes he shows exactly where agentic engineering is going and how not to get left behind. Definitely something you cannot afford to skip. Watch it, then read the step-by-step guide below on how to build a system that improves itself.

Anatoli Kopadze @AnatoliKopadze

3 months ago

x.com/i/article/2080…

108 588 4K 13.1M 18K

x.com/i/trending/208…

Today we’ve raised $52M Seed and we are announcing the public launch of S2.1 Pro. >It can clone a voice from 5 seconds of audio >2x faster than Cartesia & 1/6th the cost of Eleven Labs >most expressive model with word level control over emotion, intonation, pacing etc We support frontier AI companies including HeyGen, LiveKit, Retell, Sanas, and OpenArt all run our model in production. If you're a business and we can't cut your voice AI costs by 50%, we'll give you 1 year of Fish Audio for free. Book a demo: s.fish.audio/tmapke To celebrate our first birthday, we'll give you 1 month of S2.1 Pro for free. Like, retweet, and comment “Fish” to get it.

#PS4share