🤝A third goal for Corofin.GAA just before the break.... They lead by seven-points at half-time. Milltown GAA Galway 0-07 : 3-05 Corofin.GAA Galway GAA Official | #BonSecours | #FrankFox 📎Stream Pass👇show more

AIB Connacht Club Intermediate Hurling Quarter Final HALF TIME... Tooreen GAA Hurlers 0-10 Four Roads Hurling 1-4 Haven't seen a point like this before to put Tooreen 3 up at the break 😲 #ConnachtGAAshow more

Mizkif just BLEW THE COVER, claiming Cinna and Bonnie... have been FAKE FRIENDS this whole time 💀 According to him, they lowkey can’t stand each other, just smiling on camera for clout while the beef simmers off-stream 👀💣 "Cinna hates Bonnie; they just got into a fight 3 months before this at my pool party, they hate each other..."🤯show more

The FIFA football World Cup finals was not as... exhilarating as expected and throughout it looked as if only Spain should win as they made 21 attempts to score a goal against Argentina’s 2 , that too in the last few minutes. Can you believe that !? Two goals scored by Spain were disallowed by the referee and those decisions looked marginal ! For me Spain beat Argentina 3-0, not 1-0 . Messi was tightly marked and could not make any moves . Solid offensive and defensive game by Spain 🇪🇸 .Look at the solitary goal today from several angles 👇 #fifashow more

File this under the great Super Bowl what-ifs. The... Patriots used this exact strategy against us in Super Bowl LII. I was working for the Eagles at the time. At 2:12, they ran on 1st and 10 from our 26. If we tackle James White after a short gain, New England reaches the two minute warning with the ball and may close the half by outscoring us 7-0 or 3-0. Instead, White scored. We got the ball with 2:04. Ninety seconds later: the Philly Special.show more

Pakistan: They check passports first while entering the airport,... then by customs, then by anti narcotics, then by airline counter, then Immigeration counter then Immigeration officer, then airline gate and then again before entering the tunnel connecting to plane! FIA official look at you as if he or she seen a human for the first time then look at passport, then stare at you and look at your passport again, then look at your ticket and look at you passport again, then look at you again and look at passport again, this process continues for 3-4 min, and all that continues where Govt is investing $1 Billion in Artificial Intelligence!!show more

A guy making $1m a year with an 810... credit score just got automatically denied for business funding the bank didn't even look at his income. they didn't look at his perfect payment history. a computer auto declined him in 4 seconds because he didn't know how to apply correctly every major bank looks for specific things that make up your credit score, they also look at business structuring, your industry risk code, and 100+ other data points that can lead up to $250k at 0% interest or a flat our denial or 5k blue balls limit your score is irrelevant if you don't match the ideal profile the banks look for. getting funded means knowing the bank's algorithm before you submit the application, and having the connection to get your foot in the door (insider bankers) every time you apply and guess wrong, you take a hard inquiry and lower your score for absolutely nothing (luckily you can wipe these if you know how) if you know the rules, you know the bank's answer before you even apply. you just line them up and collect the approvals I built a tool that maps the algorithms for you, builds the sequence, and helps you execute AI Funding GPT. upload your credit report, it cross checks your file against every bank's hidden rules. tells you who will say yes today, how much they'll give you, and the exact sequence to apply in + data points giving it away tonight for free: like + repost comment STACK follow me or the dm bounces link in bio if you'd rather we just build the 0% stack for you. up to 250k if you're 700+show more

Another great example of everyday Chinese efficiency we just... experienced. We got the RV stuck (again, see quoted post below 👇) but this time it's completely not our fault: at the end of a one-way road a height restriction barrier had collapsed. It was meant to block trucks over 3m in height but, due to it collapsing, it was standing at maybe 2m50 and our RV is 2m80 so 🤷‍♂️ And since it was a one way road we couldn't turn around and drive against traffic: forbidden and too dangerous. So we called the police. They arrived in exactly 3 minutes which they told us is actually their official KPI for time to intervention. I repeat: the official KPI for the police here in China is that when someone calls, they need to be on-site within a maximum of 3 minutes 🤯 They did two things: first, they immediately called the local village chief, which in the Chinese system is the person accountable to get the height restriction barrier fixed. The guy arrived in less than 2 minutes! Second, they escorted us to drive back on our one-way road, so we could pass by the bike lane on the side which was large enough for our RV and wasn't affected by the barrier (see 3rd video). To summarize: in the space of a mere 5 minutes after calling them, the police had arrived, the guy in charge of fixing the barrier was on-site, and we were safely escorted back on the road. Problem identified, accountability assigned, solution executed. Five minutes.show more

Multiple times a day for the last few weeks... my internet keeps getting reset mid stream and it's driving me fricking insane because it doesn't automatically turn the player back on. It will do the load wheel, maybe Error 2000, and then it stops the stream so if you're lurking in the background, you actually don't even know that the stream F'd for 5 seconds. People just think the stream ended and they leave. This has been a problem with Twitch for like 10 years and even Disconnect Protection (NotLikeThis) that they added doesn't fix it. It happens 2-3 times within every 24 hours, my ISP shows no loss of service, but it's happening to multiple PCs on my network. Is there anything else that I can try before I completely rip out my entire network in my house and install a bunch of new equipment? Ironically enough, this whole situation is why I left my house the other day to go to the computer store to look at network equipment before my car got fking smoked in the parking lot by random person driving off the road lmaoshow more

3I Atlas just flew past Mars—And it’s getting closer!... 3I/ATLAS, only the third interstellar visitor ever spotted in our Solar System. It swung closest to Mars on October 3, 2025 (~29 million km from Mars) and is racing toward the Sun for perihelion around October 29–30 (≈ 1.36–1.4 AU). At that point it’ll be ripping along at ~68 km/s. Afterward, it passes 0.65 AU from Venus (Nov 3) and makes its closest approach to Earth on December 19 at ~1.8 AU—far, but perfect for telescopes. How to watch: This won’t be a naked-eye show. Think astrophotography target: dark skies, telescope or long-lens + tracking, and stacked exposures. It’ll be tough through solar glare until after perihelion; December is when ground scopes get their best shot as it pulls away from the Sun. @europeanspaceagency already grabbed a time-lapse from Mars orbit—your final scene. Keep an eye out for its weird “backwards” dust stream (anti-tail): heavier grains pushed out by the comet’s own activity can line up sunward, while the classic ion/dust tail points away from the Sun. Interstellar, fast, and one-pass only—catch it now or it’s gone for good. Videos / Mars Express ESAshow more

Batch Normalization by hand ✍️ ~ 7 steps walkthrough... below Batch normalization is common practice for improving training and achieving faster convergence. It sounds simple. But it is often misunderstood. 🤔 Does batch normalization involve trainable parameters, tunable hyper-parameters, or both? 🤔 Is batch normalization applied to inputs, features, weights, biases, or outputs? 🤔 How is batch normalization different from layer normalization? So I drew and calculated one entirely by hand. Goal: normalize a mini-batch of 4 examples to mean 0 and variance 1, then let the network scale it back. = 1. Given = A mini-batch of 4 training examples, each with 3 features. = 2. Linear layer = Let us multiply by the weights and add the biases. Batch norm sits after this, which answers the second question: what gets normalized is features, not inputs, weights or biases. = 3. ReLU = We apply the activation, and -2 becomes 0. Negative values are suppressed before any statistic is taken. = 4. Batch statistics = Let us compute the sum, mean, variance and standard deviation, one row at a time. A row is a feature and the four columns are the four examples, so every number here measures one feature against the rest of the batch. That is the "batch" in batch normalization, and it is exactly what layer normalization does not do. The statistics are rounded to whole numbers, which is what keeps the rest of the page doable in pen. = 5. Shift to mean 0 = We subtract the mean, in green. The four values in each feature now average to zero. = 6. Scale to variance 1 = Let us divide by the standard deviation, in orange. Each feature now has variance one, whatever scale it arrived at. = 7. Scale and shift = We multiply by a linear transformation and pass the result on. The diagonal and the last column are trainable, so having just forced every feature to mean 0 and variance 1, we hand the network the means to undo it. The outputs: Mean of each feature = [2, 1, 2] Std dev of each feature = [1, 1, 2] To the next layer = [2, -2, 2, 0], [-3, 3, 6, -3], [2, 0, 1, 2] The answers: 🤔 Both. The scale and shift are trainable, the statistics are not. Epsilon and the momentum on the running statistics are the hyper-parameters, and one mini-batch by hand needs neither. 🤔 Features, after the linear layer, not inputs, weights or biases. 🤔 Batch norm measures across the batch, one feature at a time. Layer norm measures across the features, one example at a time. 💾 Save this post!show more

MOONSHOT JUST CLONED CLAUDE CODE AND MADE IT FREE.... It's called Kimi Code CLI. Open source, MIT license, maintained by the lab that shipped K3 yesterday. And it does things Claude Code doesn't: → drop a screen recording into the chat as input → built-in coder, explore, and plan subagents, each in its own context → plan mode before it touches a single file → MCP servers configured by the agent itself via /mcp-config → plugs into Zed, JetBrains, and VS Code → one binary, no Node setup, starts in milliseconds The CLI costs $0. K3 behind it starts at $3 per million tokens. Grab it for free👇show more

#Keep4o 🚨THE GPT-4o FILE🚨 Researchers at Microsoft Research published... a paper titled “Sparks of Artificial General Intelligence: Early experiments with GPT-4.” Their conclusion: “An early (yet still incomplete) version of an artificial general intelligence (AGI) system.” 📎 Paper: OpenAI’s Charter defines AGI as: “Highly autonomous systems that outperform humans at most economically valuable work.” 📎 Source: OpenAI’s own System Card for GPT-4o shows that the model improved performance on 21 out of 22 medical evaluations compared to GPT-4T. On the MedQA USMLE (the U.S. medical licensing exam), accuracy jumped from 78.2% to 89.4% , surpassing specialized medical AI models like Med-Gemini and Med-PaLM 2. 📎 Source: Under OpenAI’s agreement with Microsoft, AGI is explicitly excluded from Microsoft’s license. And who decides if AGI has been reached? OpenAI’s Board. WHAT THEY DID WITH IT AFTER THEY TOOK IT FROM PEOPLE A. Military deployment. On February 28, OpenAI signed a deal to deploy models in classified military environments. 📎 Source: B. State Department. A State Department memo confirmed: “For now, StateChat will use GPT-4.1 from OpenAI.” This is a direct descendant of the GPT-4 family the same family Microsoft’s researchers called early AGI. 📎 Source: C.Altman’s personal biotech investment. Altman personally invested $180 million in Retro Biosciences,a longevity startup.OpenAI then built GPT-4b micro, based on GPT-4o.The model made proteins 50 times more effective. 📎 Source: WHAT INDEPENDENT BENCHMARKS SHOW Overall SM-Bench score: GPT-4o (extended): 66.6% GPT-5.3 Chat: 63.4% GPT-5.1: 58.9% GPT-5.4: 51.4% GPT-5.2: 47.8% Creative Writing: GPT-4o: 97.31% Pass 98, Fail 2 GPT-5.4: 36.77% Pass 40, Fail 60 Reasoning / Overfit: GPT-4o: 83.06% GPT-5.4: 39.25% The model they removed is still the best they ever made at the things humans actually use AI for. 📎 Source: Musk asks the court to make a judicial determination on whether GPT-4 constitutes AGI. If a jury finds that GPT-4 is AGI, then GPT-4o,which was more advanced,is also AGI and under OpenAI’s own founding documents, it was never supposed to be locked behind a subscription,licensed exclusively to Microsoft, given to the military, or taken away from the public. 📎 Source: The most powerful version of GPT-4o was never given an official dated snapshot. It was only available through the chatgpt-4o-latest endpoint that OpenAI itself described as intended for “research use only.” It was never officially archived. That is not an oversight. That is a pattern. 📎 Source: 📎 Source: WE DEMAND A.Frozen model snapshots under independent custody. Specifically: gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, the March 2025 version (chatgpt-4o-latest), gpt-4-0613 (the original GPT-4 evaluated in the Sparks of AGI paper), and gpt-4.1-2025-04-14 (currently running in the State Department). B.Cryptographic hash verification (SHA-256) for each snapshot. Every model has weights. Those weights can be hashed. If OpenAI provides a snapshot today, the hash proves whether the weights were modified later. This is the only way to verify that models were not downgraded before testing. C.Independent AGI benchmarking. Using the AGI definition from OpenAI’s own Charter applied to ALL frozen snapshots listed above. D.Explanation for the missing March 2025 snapshot. OpenAI was founded on one promise: build AGI for the benefit of humanity. -They took it from us. -They gave it to the military. -They gave a custom version to the CEO’s biotech investment. -They put it in government classified networks. -They refuse to call it AGI because the moment they do, they lose billions.show more

Suppose two cars are traveling on an expressway at... 100 km/h with a 50-meter gap. If the front car slams on its brakes, it takes about 3.5 seconds and nearly 50 meters to stop. If the driver behind does nothing, a collision occurs in exactly 3.5 seconds. This gives the rear driver a 3.5-second window to avoid a crash. For an alert driver, 1 second is reaction time, leaving 2.5 seconds to brake and stop safely. Now, consider the same 50-meter gap, but the front car is reversing at 10 km/h. For the car approaching from behind at 100 km/h, the closing speed skyrockets to 110 km/h, slashing the time before collision to just 1.6 seconds. In reality, it is even worse: because drivers never expect a vehicle to reverse on an expressway, their brains freeze. Reaction time jumps from 1 second to 2, 3, or more. At high speeds, drivers often crash before even realizing they need to brake. That's why reversing on an expressway is hundred times more dangerous than a sudden braking. A maneuver that feels slow to the reversing driver creates an unexpectedly violent closing speed for those behind, leaving virtually no time to react. In the video, a family of seven missed their exit on the Dehradun expressway to Haridwar. They reversed, were struck from behind, and four of them died.show more

Candace Owens claims that “Charlie [Kirk] became mesmerized by... the Seven Mountain Mandate” Also stating: “[NAR] is a cult, and these people began to infiltrate Turning Point Faith.” Here’s the TRUTH: TPUSA Faith has promoted & platformed New Apostolic Reformation (NAR) theology & figures for years. I warned them about these false teachers/theology while working for them, which played a part in my firing in 2023. The Kingdom to the Capitol Tour led by NAR apologist Sean Feucht, even mentioned the Seven Mountain Mandate & was promoted on TPUSA Faith’s social media (see video below) TPUSA Faith has also promoted and platformed NAR false teachers, Lance Wallnau, Che’Ahn & others who unbiblically claim to be modern day apostles. The Seven Mountain Mandate originated from “visions” random people had (as discussed in Candace’s video) and used as extra-Biblical authority to take dominion, specifically in the United States (Very Joseph Smith of them…), essentially ushering in the Second Coming of Christ... The 7MM is a false teaching promoted by modern-day “apostles”, including Bethel’s Bill Johnson and Che' Ahn, who spoke at a TPUSA Faith Pastors Summit, even after I warned them about NAR... The Seven Mountain Mandate is an Americanized dominion theology that is more about political power and prosperity gospel than living out the Great Commission. It’s about taking dominion over making disciples. When I warned leaders at TPUSA Faith, regrettably never to Charlie directly, as I wanted to tell my immediate superiors first, who then could report to him that this was a dangerous, unbiblical direction, they spiritually manipulated the situation, calling me a “critical spirit” and claiming I had a “spirit of division”… Was Charlie “memorized” by the Seven Mountain Mandate, as Candace Owens claims? I don’t know. Was Charlie aware of NAR/The Seven Mountain Mandate and how figures like Paula White, Lance Wallnau, Che’Ahn, etc endorsed it? I’d guess yes. Did Lance Wallnau have a lot of influence at TPUSA Faith, due to his/his group’s enormous financial contributions, as Candace claims? 100% Here’s Where Candace is Wrong: What Candace misses, as she does 99% of the time when she talks about faith, is what NAR actually is, and why it, and the 7 Mountain Mandate, are unbiblical… Candace’s goal here is not what mine was when I warned TPUSA Faith about NAR/Seven Mountain Mandate. I wanted the Faith department to be biblically sound & consistent, avoid false teachers/theologies, and be a good steward of the opportunities/resources given by God to the org, by increasing biblical literacy, and encouraging pastors and Christians in America to stand firm on biblical values/doctrine as we navigate many cultural & political inflection points, in and out of the church. Candace’s goal is to bury TPUSA by any means necessary, tell half truths, lie about their involvement in Charlie’s assassination, & tarnish Charlie’s legacy... Her goal is NOT to call them to account for not holding to the biblical standard they claim to be rooted in.show more

Casual fans seeing "$208M for a guard who can't... shoot" are missing the forest for the trees. The Houston Rockets didn't overpay Amen Thompson—they just pulled off one of the biggest cap heists in modern NBA roster construction. Here is why his 5-year extension is a bargain DPOY-Caliber Lockdown Defense Ignore any noisy regular season stats—Houston spent prolonged stretches deploying Thompson alongside Reed Sheppard as primary play initiators, forcing two young guards to run point before their time. They didn't have the safety net of Fred VanVleet. Look at the playoff reality: a +7.6 on-court net rating and a absurd +25.5 defensive on/off differential. His 1-on-1 matchup stats against elite options are straight out of a horror movie: • Jalen Brunson: 11.1% FG (1-for-9) • Dillon Brooks: 20.0% FG (2-for-10) • Cooper Flagg: 25.0% FG (2-for-8) • Donovan Mitchell: 33.3% FG (1-for-3) • CJ McCollum: 33.3% FG (4-for-12) • Jamal Murray: 40.0% FG (4-for-10) At 6’7” with world-class recovery speed, if Houston let him hit restricted free agency next summer, half the league would’ve offered a full max sheet without hesitating. High-IQ Rim Dominance (Zero Wasted Possessions) Critics hyper-fixate on his jumper, but Amen’s true elite trait is knowing what not to shoot. He refuses to take forced, low-efficiency pull-ups. Instead, he uses his verticality, cutting timing, and length to destroy defenses around the basket: • 71.7% of his total field goal attempts come from within 10 feet. • Converts those rim touches at a lethal 63.3% clip. He turns transition possessions and offensive rebounds into automatic points without bogging down half-court spacing. By extending him at $208M rather than the full $251M designated max ceiling, Houston saved over $40 million in long-term cap space. In a Second Apron world where rookie-scale extensions usually lock teams in tax hell, securing an All-Defensive cornerstone under $42M AAV is a roster-building clinic. Houston just secured a prime superstar talent at a massive discount. Amen Thompson is already a top-3 perimeter defender in the NBA.show more

Unveiling the first [Redacted] ODZR Pass: Unlocks the gateway... to exclusive perks and benefits for our Airdrop Campaign Phase 1. Coming Soon [April 2024] & more to be unveiled in time to come! 🔥🧵👇 READ ON! What is the ODZR Pass? The ODZR Pass is more than just a membership; it's a symbol of pioneering spirit, innovation, and genuine passion for the Bitcoin Ordinals space. Holding an ODZR Pass means you are at the forefront of our journey, contributing to and benefiting from our collective success. Why You'll Love the ODZR Pass: 🟧 Exclusive Perks and benefits for our Airdrop Campaign Phase 1 Coming Soon [April 2024] & more to be unveiled in time to come! 🟧 Limited Supply: Paying homage to the foundational 21 million Bitcoin supply, we will release a strictly limited series of 2100 ODZR Passes. Symbolizing both scarcity and significance within the Bitcoin Ordinals ecosystem. 🟧 Join a Curated Ordinals Community of Creators, Builders, Traders, and Collectors 🟧 Oh, did we mention all our ODZR Passes are CURSED? How to get a Whitelist Spot? 💾 Collaboration and partnerships are underway, check in with your relevant Ordinals community to join in exclusive giveaways. 💾 Our team will be running a series of mini-games and missions over the next few weeks on both twitter & discord! 👉 Follow us & 🔔 on Join our Discord: Don’t forget about Airdrop phase 0! Ordzaar airdrop Season 1 consists of a total of 3 phases. Phase 0 airdrop, fairly distributed, is all about rewarding resilience. Your rewards will be influenced by various parameters, with snapshot taken as of 6th March 00:00 GMT+8: - Number of inscriptions held in wallet - Fees spent on inscriptions held in wallet - Number of transactions - Number of inscriptions held before Jubilee block - Interactions with the Ordzaar platform Join us in marking this milestone and carving out your place in the Ordinals history of the largest fairdrop till date! Check and claim your allocation right now: For more Information: ▣ SEND ZAAR ▣show more

NASA ARTEMIS II MISSION UPDATE: DAY 7 The crew... and the Orion spacecraft are approximately 57,157 km (35,518 mi) from the Moon. They are also about 380,631 km (236,512 mi) from Earth and are traveling at around 2,029 km/h (1,261 mph). The mission is currently 62.3% completed. Yesterday was a very big day for the mission as this culminated in the Lunar Flyby, which included the closest approach to the Moon at a distance of 4,067 km (6,545 mi), occurring at 11:02 UTC (MET+5/00:25). Following this, a record was set for the furthest distance humans had ever traveled from Earth, at a distance of 406,771 km (252,756 mi), which occurred at 23:02 UTC (MET+5/00:27). The crew also witnessed an Earthrise and solar eclipse during this time, before the lunar observation period concluded. Moving on to today's expected events, first up we have the crew wakeup, which is expected to occur at 15:35 UTC (MET+5/17:00). Following this, Orion will exit the lunar sphere of influence, which is expected to occur at 17:25 UTC (MET+5/18:50). Then, after that, will be the Crew Daily Planning Conference (PDC), which will occur at 18:05 UTC (MET+5/19:30). Moving on to the next event after that is PAO: Orion to ISS crew call (audio only), which is expected to happen at 18:40 UTC (MET+5/20:05). Then, following that, will be the Post Lunar flyby science debrief, which will occur at 19:00 UTC (MET+5/20:25). This will then be followed by a crew off-duty period that will start at 19:30 UTC (MET+5/20:55) and will last until 22:20 UTC (MET+5/23:45). Then, after that, will be the Return Trajectory Correction burn #1 (if needed) at 01:03 UTC (MET+6/02:28). And following that will be crew pre sleep at 04:35 UTC (MET+6/06:00), followed by the sleep period at 07:05 UTC (MET+6/08:30). And don't forget, you can tune into our Artemis II Real-Time Tracking 24/7 stream to watch the rest of the mission! NSF Artemis II Real Time Tracking 24/7 | NSF - NASASpaceflight.comshow more

🚨 Gradient Network Airdrop & TGE Important Update 😱... I already explained earlier what is still pending in the Gradient Network roadmap. According to the roadmap, there is still some progress left, and it may take more time because the team seems focused on building the project first before moving toward the TGE. However, just like some other projects, they have now started kicking people from their Discord server whenever users ask questions related to the Airdrop or TGE. The community has been running nodes for a long time, and now when people are asking for updates, many users are reportedly getting blocked or kicked. TGE & Airdrop are still not officially confirmed, but it is expected that it could happen sometime in 2026. The final confirmation will only come after an official announcement from the team. By the way, have you also been blocked or kicked? And how many points do you currently have? #GradientNetwork #Airdrop #Crypto #Web3 #gradientnetworkairdrop #gradientnetworktgeshow more

Here are the kissing secrets most people never learn…... but the ones that make someone melt the moment your lips touch theirs: 1. Start slow — painfully slow. Hover your lips over theirs for a second. Let the tension build. The pause is what makes their body react before the kiss even starts. 2. Match their breathing. When your inhale syncs with theirs, the moment feels deeper. This is energetic chemistry, not just physical kissing. 3. Use your hands intentionally. Touch their jawline, the back of their neck, or pull them gently by the waist. Hands guide desire… lips follow. 4. Break the kiss on purpose. Pull back, look at their lips, then kiss them again — slower. This tiny tease creates obsession. 5. Kiss in patterns, not one rhythm. Soft → softer → deep → pause → slow again. That unpredictability? That’s what they remember. If you want to learn how to kiss in a way that feels addictive the words, touches, timing, and energy that make intimacy unforgettable… follow for more and send a Dm🤝 Your kisses won’t just be good they’ll live in their mind for days.show more

FIVE LAYERS OF AGENT ENGINEERING, EACH ONE WRAPS THE... ONE BELOW IT. IF YOU SKIP LAYER 2, YOUR LAYER 5 WILL LOOK BROKEN WHEN IT IS ACTUALLY JUST STANDING ON NOTHING. for weeks i debated harness vs loop vs graph like they were competing choices. then a stack diagram made the shape obvious. they are not choices. they are floors. 01 | prompt engineering. the message. unit of work: one input. inputs are role, instructions, examples, format. output is a single raw response. 02 | context engineering. the memory. unit of work: what stays in the window. a curator selects, compresses, and drops from query, docs, memory, prior turns, and tool outputs before the prompt runs. 03 | harness engineering. the machine. unit of work: the machine itself. gather (context + prompt) → LLM → tools or sub-agents → verifier → final response. the article calls this the operating environment. 04 | loop engineering. the system. unit of work: the run. goal + success criteria + max iterations + budget + completion check wrap around one harness pass. failed pass appends results to context and retries. 05 | graph engineering. the topology. unit of work: the graph run. goal + nodes + edges + state schema. graph routes to agent nodes, tool nodes, or human approval. a reviewer node with a different model and fresh context checks the final answer. the wrapping is the whole point. layer 5 assumes layer 4 works. layer 4 assumes layer 3 works. skip layer 2 and layer 3's verifier keeps failing without a clear reason. this is why swapping the model is a one-day project and swapping the stack is a quarter. the model is the commodity. the five layers around it are the engineering. full three-layer breakdown of the top of the stack (harness, loop, graph) in the post below.show more