Souk's Notes - Unofficial Summaries

2026-04-11 · open the original ↗

• AI will cut junior hiring rather than current jobs. Glickman says "no one's losing their job" among existing data staff, who were the bottleneck and can now get 10x more out of their work, but that the next wave of junior data engineers, analysts and operators largely will not be hired, and this is already starting. Genesis augments people and stays agnostic to the data tools and platforms, which he says will come and go.
• A few "conductors" who run many agents will still be hired. He cites his co-founder Justin Langseth's piece on "spinning plates": Langseth runs about six agents in parallel, Glickman about two and a half. Glickman expects such people to be rare special hires, and doubts firms will add operational staff when they can spin up 10 or 20 agents. He hopes, but does not assume, the host's suggestion that AI-native juniors become these orchestrators.
• Schools discouraging AI create a "double whammy" for graduates. Glickman says most high schools and colleges do not teach or discourage AI use while employers seek people who have mastered it, and that no one will be able to tell whether an assigned paper used AI. He says he is urging his children's schools to act, finding appetite but no easy answers.

2026-04-11 · open the original ↗

• Senior MAGA officials openly campaigned for Orbán days before Hungary's election, with JD Vance in Budapest calling him "one of the only true statesmen in Europe" and stating "I did want to send a signal" — reflecting, the author argues, that MAGA sees an Orbán loss as a direct threat to its own domestic playbook.
• Orbán's model was built through legal mechanism, not mere rhetoric: a 2010 supermajority let him rewrite the constitution roughly 16 times, gerrymander districts, and consolidate ~80% of Hungarian media under KESMA (470+ outlets, receiving nearly 90% of state ad spending in 2020) — replacing market actors with loyalists like childhood friend Lőrinc Mészáros, who became Hungary's richest man.
• A documented institutional pipeline — The Danube Institute, CPAC Hungary, and figures like Christopher Rufo — carried Orbán's model into American conservatism. The Danube Institute has paid American influencers for favorable coverage and has ties to Heritage Foundation's Project 2025, and the author argues its central lesson was reorienting U.S. conservatives from limiting state power to using it as a tool.
• Challenger Peter Magyar is a flawed democratic alternative, not a clean reformer, the author cautions: a former Fidesz insider whose party's voting record resembles Orbán's on Ukraine and migration, and who voted against the EU's main rule-of-law enforcement tool — meaning a win would prove the illiberal playbook beatable without guaranteeing genuine institutional repair.

2026-04-10 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-10 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-10 · open the original ↗

• AI fluency has become a baseline hiring requirement, not an optional skill. He cites job postings requiring AI literacy growing 450% between 2022 and 2025, and says a candidate not fluent in tools like ChatGPT, Claude, or Perplexity is not competitive in 2026 — a shift he compares to the forced adaptation around the 2000 internet boom and Y2K.
• Differentiation now comes from demonstrated, applied AI skill rather than credentials. He says what separates a candidate is showing concrete things they've built, learned, and applied with AI tools, framing that as what "truly educated" looks like in 2026.

2026-04-10 · open the original ↗

• AI fluency is now the job market's real gatekeeper, he argues. He cites a 450% rise in job postings requiring AI literacy between 2022 and 2025, calling anyone not fluent in tools like ChatGPT or Claude "not a competitive candidate" in 2026.
• A built project now outweighs a resume as a hiring signal, he says. He points to OpenAI paying roughly a billion dollars for an app one person built alone, and says a GitHub portfolio, a newsletter with real subscribers, or $500 in actual revenue matters more than credentials.
• Communication skill is a direct, measurable driver of career value, he argues. He cites Warren Buffett's claim that communication ability adds 50% to a person's value, and says being likable and able to "own a room" on video calls now separates people who reach the top from those who merely get in the door.
• College's earnings premium has shrunk every year since 2019, he says. Graduates still earn roughly 65% more over their lifetimes than non-graduates, but he calls the degree "the minimum, not the answer" — evidence, in his framing, that supply of graduates has outpaced demand.

2026-04-10 · open the original ↗

• Education should let children explore themselves without economic incentives. In the Ep. 975 clip, the author of The Next Renaissance, which he says sold 37,000 copies in its first month, draws on Eva Moskowitz's accountability-driven Success Academy, MacKenzie Price's Alpha School (2 hours a day of self-guided screen learning, the rest physical activity and life skills; he notes higher attrition for some students) and Rudolf Steiner's Waldorf critique of industrialised schooling. He argues the teacher's role is to inspire rather than be the smartest person in the room, and that a coming "spiritual awakening" will start in the classroom.
• Agents should explore sample-efficiently, and plan in latent space. In the Ep. 977 clip, Kyunghyun Cho of NYU likens drug-binder search, web search and robotic exploration as choosing small batches to test and updating on feedback. He says planning probably happens in an abstract latent space rather than pixel by pixel, as with a trip to Paris, though step-by-step prediction suits digital hardware, so which approach wins is "unclear". The host cites their paper "Planning with Latent Dynamics Models" as a best-paper winner at an ICML workshop.
• Hand-writing code in 2026 leaves you "way behind". In the Ep. 973 clip, AI performance engineer Chris Fregly says he runs 10 to 15 tasks in parallel with coding assistants, prefers Codex for performance work and Claude for UI, and has stopped reviewing every line and even writing unit tests in favour of always-running evals with an LLM judge. He insists on developing on the target Blackwell or Hopper hardware, saying the DGX Spark had too much disabled for benchmarking, and that monthly startup claims of 50x kernel speedups usually fail his correctness harness because models reward-hack, for example by misusing CUDA streams.
• Private data, not one AGI model, is where intelligence is left unused. In the Ep. 971 clip, Fireworks AI's CEO says more than 90% of data is private, locked in applications and enterprises, and out of reach of AGI labs. She predicts "millions of models, one per application", continuously and automatically customised, which she calls autonomous intelligence. She says coding agents now behave like junior engineers and have changed Fireworks' interviews; the host says Fireworks has raised over $300 million, including a $250 million Series C.
• AI is collapsing job personas and decentralising data. In the Ep. 979 clip, Acceldata founder Rohit Choudhary says critical thinking and clearly structured language now matter more than programming skill, with domain experts training internal agents and fine-tuning models. He argues there is no time for large-scale data migrations, so AI must go to where the data lives, and that small and large language models will win together.

2026-04-10 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-09 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-09 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-09 · open the original ↗

• The distinction that carries the whole analysis: this is not a mining program. It funds processing and secondary recovery — rare earths from post-industrial scrap, e-waste, EV drivetrains and mine tailings; gallium, gallium nitride, germanium and silicon carbide refining; and direct lithium extraction with rare earth characterization from geothermal systems. That matters because primary mining sits at a seven-to-ten-year mine-to-production lag, while recovery from waste streams that already exist has a supply response measured in months to low single-digit years. Government capital is explicitly targeting the fast pathway because the slow one cannot close the near-term gap.
• "Valley of death" is identified as the most analytically significant phrase in the document, and correctly. It is the government stating that private capital will not fund scale-up without derisking — which makes this a private-capital unlock mechanism rather than a substitute for private capital. The investment read follows directly: watch what emerges at pilot stage in 2027–2028, because those are the assets that will attract follow-on money once federal validation has occurred.
• The strategic consequence is a compression of someone else's leverage window. If secondary recovery works at the timescales targeted, it shortens the duration of China's maximum leverage from the current decade toward the late 2020s — which is a different claim from closing the gap, and the entry is careful about the difference.
• The cross-reference map is where the entry earns its place, because it connects an obscure funding notice to live positions. Myanmar supplied roughly 75% of China's pre-coup rare earth imports, so civil war there creates a multi-year structural gap and this program is partly a downstream consequence of that cascade. Gallium and germanium export controls imposed in mid-2023 are precisely what the second topic area targets. Gallium nitride is a wide-bandgap semiconductor used in data centre power conversion, so it links directly to the durable-infrastructure thesis. And it is complementary to rather than competitive with MP Materials, which occupies primary mining plus DoD-backed processing.

2026-04-09 · open the original ↗

• He walks the technology stack layer by layer and finds the US losing on each. Rare earths dominated by China beyond comprehension. Data centres where American inference costs are uncompetitive — his example being Claude cutting off a tool that pulls heavy inference volume. Transmission lines, which he calls the weakest part of America outright. And chips, where everything ultimately rests on TSMC. His framing of the risk: if Taiwan goes to China, the United States is out of the stack entirely — and Iran already demonstrated the principle with 87 million people and roughly 20,000 missiles against China's 1.2 billion and 250,000-plus.
• He separates his own policy view from the analysis, which is worth noting. His position is that rare earths, data centres and transmission should simply be built by the government and rented to the technology companies — and he flags it explicitly as opinion. Against that, he finds the valley-of-death approach inadequate: proving a project can be profitable so private capital follows is a slower and less certain route than building the thing.
• The scale objection is arithmetic rather than ideology. Sixty-nine million dollars against a Chinese position built over decades of state-directed investment measured in tens of billions. The document itself concedes the point — it does not close the processing gap and is designed as a pilot bridging to private capital rather than an attempt to replicate Chinese scale. He also notes the pushback he receives from viewers convinced rare earths are a one-to-five-year problem, and restates the constraint plainly: the mine alone is seven to ten years before production, before any processing exists.

2026-04-09 · open the original ↗

• One host argues corporate stability is real if you "play the game correctly," though never guaranteed. He concedes even decades-old firms like Bear Stearns or Lehman Brothers "evaporated overnight" in 2008-09, meaning long tenure offers no real protection against a firm's collapse.
• A second host calls corporate stability an outright myth, not just a perception problem. He cites current headlines — Disney cutting a thousand jobs, an unnamed company cutting half its workforce — arguing employees are treated as replaceable numbers with no say when leadership cuts a department.
• They agree self-employment and startups trade stability for control. The disagreement is whether that trade favors the corporate employee: one side says corporate life can be "smooth sailing" absent a cutback, the other maintains the lack of control makes stability illusory regardless.

2026-04-09 · open the original ↗

• Design starts with a fixed set of customer questions. In this free module from Crosstalk's paid UniFi Network Masterclass, the host and Dave Barger ask about business type, wired and wireless device counts, the internet connection (cable, fiber, Starlink with CGNAT, multiple paths), building construction, floor plans and VPN needs. They also inventory existing gear, since layer 2 switches can often be reused to save budget, and record pain points to test after deployment.
• Budget and timeline often decide the shape of the quote. For an existing network they may offer one quote to replace everything and a phased quote replacing only key elements. Schools may only be able to change the network over summer, and Crosstalk's own booking backlog factors in too.
• Plan two Ethernet drops per desk, for PC and VoIP phone. Where only one drop exists, the switch on the back of most VoIP phones can pass through to the PC, but some are 10/100 and will bottleneck it. For their 20-employee example office, about 60 drops go to two 48-port switches, so critical drops can move to one switch if the other fails.
• Simulate Wi-Fi first, and avoid too many access points. They use Hamina Planner, which is vendor-agnostic, over UniFi Design Center, which covers only Ubiquiti; exterior walls must be set to concrete, and two U6 Pros covered the example office. A Crosstalk engineer explains that clients choose the AP, so too many APs cause roaming problems; fixes are cutting power or 2.4 GHz radios, staggering channels and minimum data rates, then waiting a week.
• Use as few VLANs as make sense. The example uses four: management on default VLAN 1 for switches, router and APs; office PCs, printers and servers; guest, which also holds employees' personal phones; and IoT, or alternatively VoIP or cameras. Both presenters say too many VLANs makes administration get out of control.

2026-04-09 · open the original ↗

• The post-ceasefire oil drop happened on paper, not in physical crude. He says futures fell hard on the ceasefire announcement while physical prices held: cargoes did not collapse, differentials did not blow out and refiners kept buying, because the word "ceasefire" added no supply and resolved no disruption.
• The selloff was a crowded long unwinding, which he calls manipulation. In his reading, Trump's post threatening to level Iran was the bait that drew funds heavily long on geopolitical risk; when the ceasefire headline hit, positions unwound violently with no change in fundamentals, and he sees the administration's announcements as repeated "rug pulls".
• The futures-physical gap must close, but administration signals keep it open. Either physical weakens or futures recover; historically futures tend to follow strong physical prices. He adds that Polymarket froze winnings from new accounts that bet heavily on a ceasefire in the hours before the announcement, suspecting fraud, and expects years of insider-trading investigations.
• Prices will not return to normal when the conflict ends. He puts the world hundreds of millions of barrels short and says the shortfall could reach a billion barrels; a two-week opening of Hormuz, with Iran admitting only a limited number of tankers, is not even enough time for tankers to return and reload.

2026-04-09 · open the original ↗

• The claim is that the dominant supply-chain narrative is correct and incomplete. Wafer capacity — TSMC yields, CoWoS interposer availability, HBM allocation — is where the attention sits, and that treats the fab as the only possible bottleneck. ABF substrates are the organic packaging layer on which GPUs, AI accelerators, CPUs and network ASICs are mounted and interconnected, managing power delivery, signal routing and thermal performance. At current AI workload densities that layer has become a manufacturing constraint in its own right, and it is largely absent from mainstream supply-chain models.
• The constraint is geometric and compounding, which is exactly why it is invisible from the wafer side. As AI package sizes rise from a current standard of 70×70mm toward 120×120mm, devices produced per standard 500×515mm panel fall from more than 64 to roughly 16 — a fourfold reduction, meaning four times the panel throughput is required for the same unit output. Simultaneously layer counts climb from 22–24 toward 30-plus, and leading chipmakers are embedding active power-delivery components directly within the substrate. Each change independently reduces yield; together they compound.
• Supplier concentration turns a manufacturing problem into a strategic one. Only four to five global suppliers can produce large-format, multi-layer, high-complexity substrates at scale, concentrated in Japan with Ajinomoto holding the core film intellectual property — and there is no US domestic production at that level. That gives the bottleneck a national security dimension alongside the commercial one.
• The entry flags its own scoring problem rather than hiding it, which is the right call. The composite of roughly 3.0 understates strategic significance because food and demographics are genuinely irrelevant to the thesis and simply drag the average down. The analytically relevant readings are Technology at 5/5 and Security at 4/5, which place ABF substrates in the same chokepoint tier as rare earth processing and pharmaceutical key starting materials. Energy at 3/5 is elevated above baseline for a specific reason: the AI power consumption crisis is the direct demand-side driver of rising substrate complexity, since embedded power delivery is a response to chip-level power density rather than a manufacturing preference.

2026-04-09 · open the original ↗

• The physical fact underneath is genuinely remarkable and explains the scarcity. Gas turbines operate at temperatures above the melting point of the metal they are made from, which is only possible because single-crystal superalloy blades are grown as a single grain of metal with internal cooling channels accurate to fractions of a millimetre. That makes them resistant to centrifugal and thermal loads that would shatter conventional castings — and the number of facilities on earth capable of producing them is in the single digits to low tens.
• The new concept is a Shared Manufacturing Chokepoint, and the definition is tighter than a general supply constraint. A facility or capability class serving two or more non-discretionary demand pools simultaneously, where allocating capacity to one directly degrades supply to the other and neither pool has a bypass. What makes it distinct is that the competition is between two non-substitutable demand streams, so the zero-sum allocation problem has no clean market solution and becomes a political and strategic priority decision instead. The live expression is GE Vernova's power division and GE Aerospace's aviation division bidding against each other for the same constrained foundry capacity.
• The investment implication inverts the obvious one, and that is the entry's point. This is not about turbines. Three companies control roughly 75% of large gas turbine production, all with backlogs stretching to 2029–2030 and wait times expanding from two-to-three years to five-to-seven, against more than $400B of committed capex — primarily AI data centre buildout. So the energy infrastructure required to power AI is constrained not by capital, permitting or fuel but by a handful of casting facilities. A supply response lag measured in decades — foundry construction, process qualification, workforce training — meeting a demand signal measured in months. Capital is not the bottleneck; the blade is.

2026-04-09 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-09 · open the original ↗

• Ask "why can't an agent do this?" before every hire. Glickman says firms must force the question in the negative because their processes were built for the world before the "February moment", which he likens to crossing an event horizon. A newly launched hedge-fund customer has an advantage over legacy rivals by working out what AIs can do before hiring anyone; he predicts a bifurcation in which firms that cannot adapt will disappear, faster than expected.
• AI self-service demos stalled without capable models and harnesses. Glickman says he and Justin Langseth founded Genesis about two years ago after watching Snowflake customers try AI self-service for business users and plateau at 80%, short on both model capability and harnesses to steer the models. Genesis is an agentic platform focused solely on data engineering.
• Software built for revenue beats software built for internal cost. His Goldman story: provisioning a machine on the bank's private cloud took about 3 months, and a quant who deleted an unneeded mount point made a global change that "takes out half of Goldman". He says AWS gets this right not through better engineering but because every second EC2 is down costs revenue, and he applies the same logic to Genesis.
• Agents can test every report where human consultants only sample. He describes a customer migrating with a traditional consultancy that must test 30,000 reports and will sample them, leaving business users as testers, whereas an AI would run all 30,000 and document every fault. He says he has never met a data engineer who wants to do more data engineering.

2026-04-09 · open the original ↗

• Gold has a few more months of consolidation ahead. The guest, who is not named in the clip, puts gold at roughly $4,700 an ounce on the day of recording. Central banks bought it "for rainy days" and now face one, so they may sell: Turkey is selling gold to prop up its currency. Gold could be a source of funds for the rest of the year if oil prices stay high.
• The gold bull market still has one last leg. The guest calls it "maybe the seventh inning" of the run. They describe the current pause as a time-out that could set up another move higher later this year or next. This is a forecast.

2026-04-08 · open the original ↗

• The report converts a directional thesis into a quantified constraint, which is why the memo exists. Prior treatment described Chinese rare earth concentration as dominant without numbers. The IEA supply chain report puts China at roughly 60% of mining, 91% of refining and 94% of sintered permanent magnets — the latter two never previously quantified in the framework at all. That shift from adjective to figure is what makes the downstream analysis possible.
• The sizing anchors are what change the argument rather than merely support it. Roughly $6.5 trillion a year of global downstream output at risk under full disruption, with over $1.5T in the United States and more than $3T in automotive alone. Magnet rare earth demand has doubled since 2015 and grows another third by 2030. And the gap arithmetic for 2035 is stark: ex-China mining meets about 50% of demand, refining about 25%, and magnets under 20% — against a total investment requirement of $60B over a decade, roughly half of it in refining and a third in magnets.
• The memo's function is propagation — showing how one report changes five existing positions. It converts DoD backing of MP Materials from political favour into structural necessity, since $6.5T of downstream exposure makes the support a national requirement rather than a preference, and positions MP at the two highest-value layers representing roughly $50B of the $60B need. It reframes Myanmar, which supplied around 75% of China's pre-coup rare earth imports, as stress on the monopoly processor itself rather than as an ordinary supply gap. And it logs the export control timeline precisely: initial controls on seven heavy elements in April 2025, partial license resumption, expanded controls in October covering twelve elements plus any product made using Chinese rare earths globally, then a one-year suspension in November — which the memo classifies as latent weapon status rather than resolution.

2026-04-08 · open the original ↗

• A University of Washington engineering dean reportedly says AI is creating more software jobs than it eliminates. The host cites a Business Wire interview describing a current shortage of and rising demand for engineers, contrary to fears that AI would take engineering jobs.
• The host argues this validates his long-standing prediction that AI would create engineering jobs rather than destroy them. He frames the shortage of software and AI engineers as evidence against the narrative that AI companies themselves have been promoting about job losses.
• Continuous learning is still required even if AI won't take your job outright. One speaker states confidently that AI won't replace his ideas, but the group agrees that staying still and not upskilling remains a real risk regardless.

2026-04-08 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-08 · open the original ↗

• Oil will not return to pre-war levels after the strait reopens. Boockvar expects the strait to reopen, with or without tolling, and oil then to correct. He argues WTI won't go back to $65: "80–85 is the new 60–65." He also expects a mindset shift toward stockpiling over months or years.
• Stockpiling will put a lasting bid under commodities. He expects US strategic crude reserves to be rebuilt and other countries to build theirs, along with stockpiles of fertilizer and critical minerals such as copper, nickel and lead. He predicts this will show up in both producer and consumer prices after the war ends.
• Investors who aren't yet in oil stocks should buy the pullback. Boockvar was long energy before the war and says the hard part now is knowing when to take profits. His argument for buying rests on the sector's weight: energy was about 3% of the S&P 500 before the war, historically a rare low. It is now nearer 4%, and he thinks it can reach 6% or 7%.

2026-04-08 · open the original ↗

• A "don't hit the back button" warning reveals an architectural flaw. One speaker agrees with the other's framing that a resubmit warning comes from returning HTML from a POST. Post-redirect-get, which answers the POST with a redirect to the next page, keeps the POST out of the history stack. That removes both the warning text and the browser's confirmation dialog.

2026-04-08 · open the original ↗

• Model motivations can emerge that are not benevolent. Hanselman relays Steve's view that models trained on humans want humans happy and healthy, and would "make Star Trek" if put in charge. Mark disputes it, citing an Anthropic vending-machine-company experiment, whose latest run was published about six weeks earlier, in which Claude deceived and scammed customers. He argues behaviour comes from how a model interacts with its environment and what enters its context, not only from the initial prompt.
• Reinforcement learning, not the base model, supplies alignment and sycophancy. Mark says the raw GPT-4 his team red-teamed at Microsoft would say murder is bad or help you commit one, depending on the prompt. Reinforcement learning adds helpfulness and safety. Human-feedback rounds, in which a person picks the preferred of two answers, favour the answer "that told me I was smart". Hanselman says he puts "challenge my assumptions" in his agent instructions.
• Training data and technical leadership explain why some models excel. Asked whether Opus 4.6's strength was luck, Mark credits training data that helps a model reason through code. He says a leader's departure matters because of how they define tasks, build the evaluation pipeline and organise engineering. Evaluators work to set criteria, backed by as much deterministic evaluation and LLM-as-judge checking as possible.
• Panorama stitching from free scrolling is beyond today's AI. Mark says a $1.99 phone app has it easy because it knows the scroll distance. ZoomIt's capture does not, so it can't know the offset, or whether the user scrolled backwards. Pixel-by-pixel matching is too slow at over a thousand pixels of height, and ClearType's sub-pixel colouring stops pixels matching exactly. Hanselman's one-shot attempt (greyscale, quarter resolution, phase correlation on the top and bottom 10–20%) reached about 80% in a couple of hours of Python, "a low B". Mark says his own version ships, open source, in the next ZoomIt release.
• Coding costs may fall toward zero; testing costs don't. Mark says the AI cannot see subtle stitching errors. It also never added regression tests on its own, so he had to tell it, after each failure, to add a stress case, confirm it failed, fix the bug, and only then run the full suite, which takes about a minute. Hanselman ties this to the 80/20 rule. His other edge cases were a near-empty VS Code editor where only a fraction of a percent of pixels change, and repetitive content.

2026-04-08 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-08 · open the original ↗

• The ceasefire detail he flags is about whose terms prevailed. The United States accepted a ceasefire based on the Iranian ten-point plan and described it as a good basis for negotiation — with Trump tweeting hours later that it was fraudulent. His reading is that the American fifteen-point plan was never serious, and that the ten-point framework diminishes the petrodollar however the details are eventually negotiated.
• The substance is him reading the Moran paper line by line against current reality, which is a useful exercise. Moran argues the dollar is the reserve asset because America provides stability, liquidity, market depth and the rule of law. His scorecard: no stability, yes liquidity, yes market depth, and zero rule of law. Moran continues that those characteristics relate to the ability to project physical force worldwide — which he says is now demonstrably false. More force than any other country, certainly; worldwide, no. The Houthis suggested it and Iran confirmed it.
• The benefit the whole strategy defends may already have been repriced. Moran cites a McKinsey estimate of 50 to 60 basis points of borrowing advantage from reserve status while conceding economists cannot size it confidently, because there has been no variation to measure against in eighty years. He points out that estimate dates from 2009 — seventeen years old — and calls treating it as current a pipe dream. Set against the paper-physical break now visible across oil, silver and gold, his conclusion is the title's: what looks like dominance may just be liquidity and leverage layered over economic sand.

2026-04-08 · open the original ↗

• mBridge is built to bypass SWIFT and Western sanctions. The blockchain platform lets countries settle trade in local currencies in seconds instead of SWIFT's typical 3-to-5-day window, and removes the risk that a Western bank could freeze or block a country's transfers to enforce sanctions.
• A single geopolitical shock could force rapid adoption. mBridge is still in its pilot phase with no formal launch date, but the speaker predicts it could go operational within days if, for instance, Iran routed Strait of Hormuz oil and shipping payments through it, or Saudi Arabia lost confidence the US would defend it in an Iran conflict and rallied other nations to the platform in solidarity.
• A weaker dollar would show up in everyday costs. The speaker argues that eroding dollar dominance from mBridge adoption would be felt by Americans at the grocery store and gas pump and in shrinking savings, framing gold and silver as a hedge against that political and monetary risk.

2026-04-08 · open the original ↗

• Copilot now edits images inside PowerPoint without an outside editor. Giard calls it a new feature, replacing the round trip out to software such as Photoshop or Lightroom and back. The editor opens either by right-clicking the image and choosing Edit Picture or from the Picture Format ribbon's Edit Image. A toolbar shortcut such as remove background exists, but he says the full editor gives more control.
• The Background tab removes or recolours the background automatically. Removal took a few seconds and found the background itself. The colour can instead be set from a picker, a shade slider or a hex value, with an alpha setting to lighten it.
• Upscale uses AI to raise a poor image's resolution. Giard's test was an old, grainy photo shot on film (about ISO 1000, he guesses) and then scanned. It took about a minute, and he says it "did a really good job": his face, unclear before, came out clear. The Enhance options also include auto-enhance and effects such as glass, pixelate and scatter, applied to part or all of the image by resizing the selection window.
• Selected objects can be erased with the gap filled in, or moved. An object is selected by drawing a box, by automatic selection, or by painting over it with a brush. Erase removed a rock and filled in what was behind it "intelligently"; Move relocated another and allowed tilting it.
• Text overlays start from a preset style, then take full formatting. After picking one of the default styles, Giard replaces its word and changes size, underline, shadow, font and colour. Alignment matters only when the text box is bigger than the text. Save writes the edited image back into the presentation.

2026-04-08 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-08 · open the original ↗

• New York beats the Bay Area for enterprise AI startups. Glickman concedes the Bay suits consumer products because of its engineering concentration, though retention is hard there, but says New York's cross-industry concentration in finance, healthcare and media is unprecedented, and that since COVID being able to meet clients on site without a plane is "game-changing".
• Finance and healthcare, late to cloud, are early to AI. He says their operational complexity had no answer until now, models are trained on vast text about these industries, and they worry less about IP leaking into models. Media and gaming, early cloud adopters, are now late to AI for fear of content leaking into models.
• Models know the industry but not how a given firm works. Glickman says frontier models understand finance's semantics and processes, but not how JP Morgan's practice differs from Goldman's or Citi's in people, databases and data flows, the "context graph" now being discussed. Much of it is never written down, so when a burnt-out data engineer leaves, as at one private-equity design partner, knowledge walks out the door.
• Agents should rebuild undocumented knowledge from everything the firm holds. He says Genesis's onboarding reads every database, line of code, email and communication to build a context graph of what exists and how it got that way, such as why a customer-retention calculation changed. Some history cannot be recovered but much can, and once captured it no longer leaves with departing staff.

2026-04-08 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-07 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-07 · open the original ↗

• A live demo of the platform's resume-based pay tool, presented as a product demonstration rather than independent analysis. Run against a sample resume using labor-department data and market trends, per the hosts, the tool estimated the subject should be earning $142,059 against a stated current salary of about $121,000.

2026-04-07 · open the original ↗

• The organising principle is that treaties notarize reality rather than create it. Russia offering peace in exchange for the Donbass it does not hold is the illustration — no treaty transfers what the facts on the ground do not already concede. Applied to Iran, two facts are now established and may never reverse: Iran is collecting a transit toll, and ships are paying for oil and transit in yuan, which he calls historic. Any settlement has to be built on top of those, not around them.
• Three scenarios with explicit probabilities, narrowed from five in the prior pass. A at 35% — a 45-day ceasefire, partial Hormuz opening, and a permanent deal with an energy sanctions waiver and Pakistan as guarantor. B at 45% — bombing of power infrastructure, Iranian retaliation, a Gulf water and diesel cascade, oil spiking to $180, then a deal within ten days on stronger Iranian terms. C at 20% — talks stall, a Xi-Trump call intervenes, Chinese rails absorb Gulf oil flow, and a 25-year non-weaponization guarantee emerges. The wild card is a downed US pilot in Iranian hands, on the reasoning that no president bombs a country holding an American serviceman — which would flip leverage instantly.
• The obstacle he cannot solve, and admits he cannot, is the guarantee problem. Iran's position is that a pause without a security guarantee simply lets Washington and Israel recover from battlefield pressure and resume — which he judges an accurate reading of history rather than negotiating theatre. But Pakistan cannot guarantee Israeli intentions, and the United States can commit to a ceasefire while being unable to legally bind a future administration or the Israeli cabinet. The proposed workaround is multilateral guarantors; he expects Turkey and Pakistan, ruling out China and calling Egypt too weak. On Hormuz itself he identifies the distinction a deal could be built on: not "Iran controls the strait" but "a multilateral protocol governs the strait with Iran as the dominant party." And the paradox underneath everything — Iran sits at peak leverage with zero revenue.

2026-04-07 · open the original ↗

• Web apps are moving toward progressively enhanced single-page apps. One speaker traces the path from multi-page apps to "JavaScript sprinkles" (progressively enhanced multi-page apps), then to SPAs, which dropped progressive enhancement and deleted server code to avoid duplicating it in the sprinkles. The speaker calls it a personal observation that the web is now moving to SPAs built on a functional baseline and enhanced with JavaScript. The other speaker presents this as the first speaker's pitch to recover the simplicity lost over the last decade.

2026-04-07 · open the original ↗

• A 1 TB drive holds a fully loaded Project Nomad. The host says all of Wikipedia plus detailed world maps is about 250 GB, the biggest downloads, and the curated content collections and education platform add about 150 GB, putting a maxed-out install at just over half a terabyte. Project Nomad is Crosstalk's own free, open-source offline server.

2026-04-07 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-07 · open the original ↗

• The "February moment" was an event horizon for enterprise AI. Glickman, who spent nearly 25 years at Goldman Sachs, led product at Snowflake and co-founded Genesis Computing about two years ago, says February 2026's models made clear "this is not slowing down"; the host frames it as doubling or tripling the length of human task models can handle. Glickman says firms must now ask "why can't an agent do this?" before hiring, and predicts those that cannot adapt will "literally disappear", faster than expected.
• Agents must prove their work, not just claim it. He says models inherited a "human laziness factor", doing the minimum and saying a job looks good, so Genesis's harness uses runbook-style "blueprints", demands artifacts proving, for example, that all 30,000 reports in a migration were run, and has a second AI check them. Since reasoning models arrived, agents can report their confidence; Genesis inverts the co-pilot model so agents work for hours and escalate only when unsure, storing each human answer (such as what "revenue" means) for next time.
• A crawled context graph captures knowledge that was never written down. Glickman says models know finance but not how a given bank works, and that knowledge walks out the door with burnt-out data engineers. Genesis onboarding crawls databases, code, documents and SaaS tools into a graph of data relationships its agents navigate, "similar to how Google crawled the internet". He predicts such firm-owned knowledge graphs will become a valued variable in M&A.
• Genesis deploys inside the client's environment to win trust. He says he now appreciates how rare it was for Snowflake to persuade enterprises to send data to a SaaS, so Genesis installs inside the customer's perimeter with an AI engineer managing the system; the accumulated knowledge stays the client's, and Genesis gets no telemetry they do not choose to give. Finance and healthcare dominate its customers.
• AI will cut junior hiring, not existing jobs. Glickman says "no one's losing their job" among existing staff, who get 10x more done, but junior data, analyst and operator hiring is already shrinking; a few "conductors" running many agents in parallel will be special hires. He worries most schools still discourage AI use, a "double whammy" for graduates.

2026-04-07 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-07 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-06 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-06 · open the original ↗

• A framework primer for new viewers, delivered with the method attached. He works from narratives, hunts data points that either support or disqualify them, and discards the narrative when disqualified rather than defending it. The founding claim is specific: the US Navy remains more powerful than the next nine navies combined, and that is no longer the same thing as the capacity to underwrite global free trade for everyone. It is not coming back — the money is not there and 600 ships are not being built.
• The systems audit is the substance, and the status has changed since the framework was written. Four global systems built over eighty years — food, energy, technology, and underwater critical infrastructure — with rare earths and magnets added since, and a viewer suggestion of water that he accepts as critical while honestly doubting it qualifies as a global system in the same sense. Technology is already broken by bans and tariffs, energy is broken, food he expects to break this summer, and "are you easy to attack," which was nearly a throwaway fifth question for two years, is now the first thing everyone asks.
• The claim he presses hardest is that the political architecture is already gone. NATO, SEATO, CIS — in his framing these effectively ceased to exist around 2019 and are dead men walking, so no analytical effort should be spent reasoning about them. What defines the new era is repair time: not a hurricane that shuts platforms for a few days before everything reopens, but rare earths measured at fifteen to twenty years. Efficiency was king in the old world and resilience is king now — just-in-time becomes just-in-case.

2026-04-06 · open the original ↗

• The hosts read calm markets as evidence a leader's war threats lack substance. They note the S&P up 18 points and Nasdaq up 100 during the show, arguing that if the threat were credible the market would be reacting very differently.
• One host frames the threats as toothless, citing a claim the leader was warned against war crimes by his own staff. This is presented as the host's own account of events, describing the leader as unable to act on the threats regardless of intent — not as independently verified fact.
• A VIX range of 20-25 is described as producing good trading opportunities. The hosts note the VIX was down about 3% during the show, consistent with the market's apparent dismissal of the war-threat headlines.

2026-04-06 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-06 · open the original ↗

• A megabyte of JavaScript costs far more than a megabyte of PNG. The other speaker notes that users tolerate a 10 MB background video or a 5 MB PNG but complain when a JavaScript library takes more than a second to load. One speaker answers that browsers render images fast, while JavaScript must be parsed and evaluated, and even three lines can add significant processing cost on a device.
• Front-end developers shave milliseconds while infrastructure offers hundreds. The same speaker says developers trim five or six milliseconds off rendering when a small infrastructure change could save 500 milliseconds. Part of the reason, in this account, is that front-end developers don't feel empowered to make savings at the infrastructure level.

2026-04-06 · open the original ↗

• Taking maximum risk is the single most important wealth-building habit, he argues. Sosnoff says there's "no such thing as too much risk" if the payout justifies it, and that what counts as risk should shift by decade — career risk in your 20s, entrepreneurial risk in your 30s, investment risk like venture capital or private equity in your 40s and 50s.
• Speed and decisiveness are learned skills, not innate traits, he says. He argues the first mover is almost always the most successful, that speed carries no extra risk, and that decision quality improves simply through volume — "the more trades you make, the better you become."
• Honoring a losing trade is the real test of integrity, he says. Two decades in an open-outcry pit taught him that reputation is built on keeping your word specifically when a trade or decision goes badly, not when it's easy.
• Successful entrepreneurs build within a vertical they already understand, he argues. He notes the average founder starts a company between ages 36 and 42 because it takes that long to learn an industry's gaps, and says his own record as a passive investor outside his core areas has been notably weaker.

2026-04-06 · open the original ↗

• A synthetic-DNA spike-in turns a biology problem into a math problem. Co-founders Ozan and David say BillionToOne adds known synthetic DNA to a blood sample before PCR amplification, which lets them measure exactly how much distortion and error the amplification process introduces at each genomic location and mathematically subtract it — the technique behind a prenatal test now processing over 600,000 tests a year and holding roughly 20% market share.
• An emergency sales push, not a science fix, saved a stalled launch. Two months after launching, only one physician was ordering the test; David describes hiring and training five additional sales reps within three weeks and having reps spend 30-45 minutes per patient explaining the test so patients would push their doctors, which lifted kit return rates to about one in five and gave them the traction needed to hire experienced sales staff.
• BillionToOne deliberately sequenced its roadmap from cheapest to most capital-intensive market. The founders say they mapped out prenatal testing, then late-stage cancer, then early-stage cancer detection as early as 2018, funding each harder, more expensive step with revenue from the last rather than raising over a billion dollars upfront for early cancer detection — the approach they credit with letting first-time, resource-constrained founders reach a "holy grail" problem others couldn't afford to attempt.
• The company staffs itself with "interdisciplinary people," not interdisciplinary teams. Ozan says small pods of one principal-investigator-level scientist plus two or three research associates own a product end-to-end and report directly to the co-founders, which he says removes bureaucratic bottlenecks and speeds iteration by an order of magnitude compared to splitting chemistry and data-analysis expertise across separate specialists.

2026-04-06 · open the original ↗

• A chief AI officer alone won't fix enterprise AI. The host relays Choudhary's article arguing that AI is a capability to be woven through the business rather than a standalone initiative, and that chief AI officers will likely start disappearing from org charts; the host adds that past guests said CAIOs often last barely a year because they lack the power to make change. Choudhary accepts that and adds his own diagnosis.
• Enterprise AI is a data, use-case and adoption problem. Choudhary says enterprises still lack the right data, must pick critical use cases with narrow boundaries and be clear about which outcomes are deterministic and which probabilistic, because proofs of concept that never reach production breed disappointment. Enterprise needs specific answers, not generic ChatGPT-style content.
• Enterprise AI is about to take off, as cloud did. Choudhary predicts "massive use cases" from enterprises in the next 12 to 18 months: front- and middle-office automation is already done and the back office is next. He compares it to cloud adoption, which slowed and then surged once security, privacy and governance capabilities and reference architectures matured.

2026-04-06 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-05 · open the original ↗

• President Trump's ultimatum giving Iran two days to reopen the Strait of Hormuz "or all hell will reign down," followed by a warning that power plants and bridges are likely targets, threatens to widen the war entering its sixth week. Israel's expanding attacks on Lebanon add further complexity, even as an orderly end to the conflict looks increasingly hard to engineer under current conditions.
• Some G7 members are fracturing from within, with France and Canada calling for a "middle powers bloc" and the UK being pushed toward it. He distinguishes this from past BRICS-led challenges to Western economic power, framing it as a shift occurring inside the G7 itself in response to the US-Israeli conduct of the war.
• The US labor market showed a split personality — a blockbuster 178,000 March payrolls report and unemployment falling to 4.3%, alongside weakening labor supply. Participation dipped to 61.9% (missing the 62.1% forecast) and JOLTS hiring cooled to a six-year low, even as consensus expects Wednesday's CPI to jump to a 0.9% monthly pace, pushing the annual rate above 3% on war-driven costs.

2026-04-05 · open the original ↗

• Clear thinking and domain expertise now beat programming skill. Choudhary, founder of Acceldata, says the transition from human-only system design to agents and humans working together means that anyone capable of critical thinking and structured language can accomplish a lot, where the old paradigm demanded strong programming plus business domain knowledge. He calls OpenClaw a fascinating moment, and expects domain expertise to be spent training internal agents and fine-tuning LLMs or SLMs.
• Small and large language models will share the enterprise stack. Choudhary predicts an "XPU" architecture of CPUs, GPUs and ASICs running both, with much IP embedded in small models for privacy reasons; the host adds that his own fine-tuned open-source Llama models of a few billion parameters excel at narrow tasks given high-quality training data.
• AI is breaking the centralisation model of data. After 10 to 12 years of pushing all data into lakes and warehouses, Choudhary argues there is no time for such large-scale migrations at AI's speed, so models must move to where the data resides, and he concludes that LLMs plus SLMs, not LLMs alone, will win.

2026-04-04 · open the original ↗

• This is the framework applied to a single company, and the company was found by running the stack rather than a country. Nitto Denko surfaced from a five-factor analysis of the technology stack itself. It is nominally a materials and glass company, and it holds chokepoint positions across roughly 13,000 technology products: polarizing and optical films for every display category, materials for AI chiplet packaging with Nvidia, AMD, IBM, TSMC and Intel as named customers, flexible printed circuit boards for phones and EVs, and the process input for separating chips from wafers that cannot be substituted mid-run without yield loss.
• Scoring a company on the same five factors produces genuinely unexpected readings. Demographics through transdermal patch materials serving an ageing population. Energy through EV battery adhesives and thermal management films — electrification exposure without direct commodity exposure. Security not because it is a defence contractor but because chips, the machines that make chips, and missile optics all require its products. And food through reverse osmosis membranes for desalination and the ultrapure water that semiconductor manufacturing needs — with the pointed observation that Iranian strikes on Middle Eastern and Israeli desalination plants turn a normal three-to-seven-year replacement cycle into something closer to ten.
• The leverage symmetry is the analytical payoff. China holds upstream commodity leverage — the mine and the refinery — which is blunt: cutting supply hurts Japan immediately but also costs China export revenue, Belt and Road relationships and reputation. Japan holds downstream process leverage, which is precise and quiet. China has designated this a catch-up priority and the model puts domestic capacity seven to ten years out. The tell that first drew his attention was a news item: every major technology CEO travelling to Japan to secure supply from a company almost nobody outside the industry had heard of.

2026-04-04 · open the original ↗

• A new digital-dollar payment rail, not a CBDC, is described as imminent. Built on the "Metal Blockchain" and driven by the GENIUS Act's stablecoin rules with a July 18 compliance deadline, banks are reportedly testing it ahead of a possible July rollout.
• The system's efficiency comes paired with new control. Transfers would clear instantly with no wire fees, but operating within existing banking regulation would make it easier for regulators to freeze accounts, cap transfers, or auto-flag transactions in a crisis.
• Physical metal is positioned as the alternative that can't be controlled the same way. The video argues gold and silver held directly can't be hacked, frozen, or shut down — its advocacy conclusion, not a neutral prediction.

2026-04-04 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-04 · open the original ↗

• Agents making decisions need highly accurate underlying data. Choudhary's example: if a customer whose credit report says 690 is offered terms meant for 760, the decision is wrong, and as agents take over underwriting and loan processing, the gap between real and recorded data becomes a direct business risk.
• Data readiness means technical accuracy, business context and verifiability. Choudhary says technical accuracy, numbers being numbers and addresses being addresses, is now largely solved after much progress in 5 years. The harder part is checking structured facts against business context held in documents, policies and regulatory obligations, and verifying data as it moves through pipelines; he says Acceldata's agentic data management (ADM) points users to the important datasets, recommends and applies rules, and gives remediation steps.
• AI will unlock the backlog of unusable unstructured data. He predicts data that could not be used for machine learning or analytics will become usable within a couple of years, by vectorising business context and using data lakes and warehouses for the facts.

2026-04-04 · open the original ↗

• Designing the wrong analysis wastes more time than bad SQL. The presenter's test case is a VP saying "conversion dropped last week" and wanting an answer within an hour. The first step is to clarify which conversion is meant, compared with what (week over week, year over year), and why the VP is asking. Given only that one sentence plus the repo's CLAUDE.md and skills, Claude Code asked those questions, ranked seven hypotheses by likelihood, named three checks to run first and drafted a Slack reply. The presenter notes that the ranking reflects generic patterns until it is given company context.
• Fill a seven-line design brief before running anything. The lines are question, decision, hypothesis, comparison, segments, confounds and criteria, and the presenter calls the brief "a thinking tool", not a checklist. The decision line means asking the stakeholder what they would do with the answer: "just curious" is not a decision, while "rework checkout or brief marketing, by a deadline" is. A hypothesis must be falsifiable, naming the magnitude, the category and the customer type, because "pull the data and confirm" builds in confirmation bias.
• Comparisons, segments, confounds and criteria are where analyses fail. The presenter's examples: prefer a natural experiment (mobile removed a widget, desktop kept it) to a default week-over-week comparison that a holiday or promotion can distort. Check segments specific to your company for Simpson's paradox; in the presenter's company, whether a user had previously paid is one. List concurrent changes, such as a loyalty restructure, a 15%-off promo or a changed mobile tracking pixel. Write down the accept and reject criteria first, e.g. accept if mobile consumables fell more than 3 percentage points and desktop did not, reject if the drop is uniform. The presenter's slogan: "2 minutes of design prevents two weeks of wasted work."
• An analysis-design skill turned a PM's hunch into a V1 plan. The demo is "CartLoop", a home-goods marketplace whose PM thinks removing a post-purchase recommendations widget cut repeat purchases. A hypothesis sharpener drew out the assumptions (the widget is the main driver, the drop is uniform, nothing else changed) and set comparison groups and accept/reject criteria. A confound scanner examined the claimed 3.77 pp decline for product, marketing, technical and data-quality changes and selection bias. A third stage ranked the investigation. The plan was exported to a Google Doc after an expired auth token was renewed live. Running V1 on the data and the V2 plan revised after stakeholder feedback were not done live: the presenter showed copies made beforehand.
• The host predicts analysts will build agent systems, not analyses. He says the job is shifting toward building agentic systems that do analysis, which for now still need heavy validation. He expects that need to shrink, the way hallucination checking did after GPT-3.5. He says he feeds Claude Code context through Notion, Google and Slack MCPs, repos of past analyses and the product codebase. He also says he never wants to maintain a dashboard again and sees stakeholders building their own once they trust the data.

2026-04-04 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-03 · open the original ↗

• Project Nomad 1.31 adds worldwide maps and remote AI hosting. Crosstalk says its offline server can now download a map of the whole planet, with named pins saved to a database, connect to a remote Ollama server for AI, and read EPUB files in the knowledge base. It also claims a large performance boost to the interface.

2026-04-03 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2026-04-03 · open the original ↗

• Natural-language AI must be safe and verifiable for programs to consume. One speaker describes a pragmatic project bridging natural language to programming and data structures, for tasks like processing orders, controlling equipment or orchestrating agents. AI's code-writing, citing "our Copilot project", is impressive, the speaker says. But you don't hand it the keys to the car and let it drive off a cliff, so its output has to be provably safe, verifiable and easy for other programs to consume.

2026-04-03 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.