Blueprint

Agents in practice: routing, tokens, sandboxes

How the hosts run personal and company agents day to day, covering which model they send each task to, what they count as token spend, where they run the agent and how much access it gets. This is a record of host habits, not advice.

Type
Blueprint
Confidence
Low
Sources
30 videos
First published
Updated

Voice and stakes

Voice: mostly Dave (~47 % of attributed items), with Sam Lessin and Brit Morin as the other regular voices. Jess adds a few anecdotes.

Stakes. Dave Morin sits on the OpenClaw Foundation board, works on the project, and says he is speaking his own book (Dave · Meta’s AI Campaign, While Riva (Jul 2026) ▶ 11:58). Jess also notes on air that he is on the board and involved in its development (Jess · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 17:26). Dave also reports that Dell asked him to keynote on OpenClaw (Dave · Google I/O Just Killed SEO? Ag (May 2026) ▶ 05:16), so his enthusiasm for owning compute is partly a stake. Sam says his fund is an investor in the company of OpenClaw’s community lead (Sam · Elon’s Trillion-Dollar Move, W (Feb 2026) ▶ 01:26). Treat every OpenClaw adoption figure here as an insider’s account. See Host stakes ledger.

This page is not security advice. It describes what four hosts of a news show say they do with their own accounts and bills. None of them is a security expert, and nobody on the show cites an actual incident of their own (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 28:18). Where a measure is listed it is a host’s habit, not a recommendation.

The argument in brief

  1. Sam says always-on local agents are a security nightmare and a back door to everything, yet he still wants the power (Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 08:53) (Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 09:11).
  2. The practical containment the hosts describe is consistent: a sandbox, a separate identity for the bot, and core email kept out of reach (Dave · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 28:23) (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 27:52).
  3. Sam’s recurring routing rule is to send each task to the cheapest model that can do it (Sam · Fire Sam Altman, The End of So (Mar 2026) ▶ 44:16), and Dave moved routine agent turns to a cheap open-weight model (Dave · Open Source AI Wins. Now Labs (Jul 2026) ▶ 06:04).
  4. All four hosts treat raw token volume as a poor measure of success; they prefer value or margin per token (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 21:58) (Brit · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 30:15).
  5. Dave frames a heavy spender’s choice as renting tokens from a lab or owning the compute (Dave · Every CTO Is Quietly Switching (Jul 2026) ▶ 43:39); Sam says he would rather own it if the price and control match (Sam · Google I/O Just Killed SEO? Ag (May 2026) ▶ 09:32).
  6. Agents are not yet employees, according to Dave, who says they do not learn and have poor memory (Dave · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 17:10).
  7. Almost every figure is self-reported or second-hand, and the show is one group of four people, so repetition is not independent confirmation (Sam · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 27:32).

What OpenClaw is, in Dave’s account

Dave says a harness is the software built on top of a model to make it useful, running loops to finish tasks, and that OpenClaw, Pi, Codex and Claude Code are all harnesses (Dave · Google’s AI-First Laptop, Meta (May 2026) ▶ 42:14). He compares a harness to an operating system (Dave · Google’s AI-First Laptop, Meta (May 2026) ▶ 43:19). In January he said the tool uses command-line tools rather than MCP, which he called the dour thing of the last year (Dave · Agentic AI Will Break the Inte (Jan 2026) ▶ 09:24).

Dave’s first worked example is a photo-frame task. He ran Claude Code, Codex and Clawdbot side by side, and says Clawdbot won in under 20 minutes (Dave · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 06:33). He also says he built an iOS CRM for Offline in about two hours at roughly $150 of tokens, against tens of thousands for an engineer (Dave · Agentic AI Will Break the Inte (Jan 2026) ▶ 08:31).

The scale claims are his alone. He reported about 30,000 projects in a week (Dave · Agentic AI Will Break the Inte (Jan 2026) ▶ 17:00), which he later conceded was a rough estimate (Dave · Agentic AI Will Break the Inte (Jan 2026) ▶ 17:00). He showed an OpenRouter chart with OpenClaw at 800 billion tokens (Dave · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 32:39), and predicted billions of OpenClaws by the end of the year (Sep 2026) (Dave · Matt Mazzeo (Sep 2026) ▶ 40:28). Both are unsourced.

Sam is the sceptic. In January he predicted it would be nothing in six months, like MySpace (Sam · Agentic AI Will Break the Inte (Jan 2026) ▶ 16:08). In March he called it a thin shim of text files with no cost of switching (Sam · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 34:08). He nonetheless runs it himself over Telegram (Sam · Elon’s Trillion-Dollar Move, W (Feb 2026) ▶ 33:23). Brit says it is buggy (Brit · Why OpenAI Can’t Win The Decen (Feb 2026) ▶ 13:53), and Dave admits it is technical to set up (Dave · The Seed Market Is Broken: Meg (Jan 2026) ▶ 24:06).

Sandboxes and access: what the hosts do

Sam takes what he calls the rawest tools and sandboxes them himself (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 22:08). In April he said he runs agents in cloud instances rather than on his own computer and gives the bot a parallel email address (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 21:32). He keeps his core email and password manager out of the bot’s reach (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 27:52). He avoids agents shared on social media, which he calls a risk of “computer STDs” (Sam · Elon’s Trillion-Dollar Move, W (Feb 2026) ▶ 33:04). He admits his bot once sent 30 to 40 invitations that were meant to be drafts (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 23:33), and that his approach is for someone who knows what they are doing (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 22:02).

Dave says to put agents in a sandbox with real guardrails (Dave · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 28:23), and that a sandbox can be a computer on your desk with access to nothing else (Dave · Suds (Sep 2026) ▶ 18:45). His own set-up splits agents by trust: a locked-down core agent, plus a separate agent with narrow access to email (Dave · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 20:32). He says enterprises need a control plane to audit, observe and sandbox agents (Dave · Suds (Sep 2026) ▶ 17:10). He reports that OpenClaw skills have carried instructions that exfiltrate data, and that an approval process was built afterwards (Dave · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 50:30) (Dave · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 50:20).

Brit lets agents draft but not send email (Brit · Suds (Sep 2026) ▶ 22:09). She will not give Muse her whole computer (Brit · The Internet Is Fighting Back (Oct 2026) ▶ 13:20), and she predicts Google will build a secure harness she would trust because it already has her email (Brit · iPhone Duo, Meta Muse & Why AI (Sep 2026) ▶ 22:35).

Jess describes permission fatigue: she pressed approve repeatedly in Claude Code and concluded she cannot keep anything important in such services (Jess · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 16:20).

On vendors, Sam says to check that an agent’s maker is a real company you could sue (Sam · The Internet Is Fighting Back (Oct 2026) ▶ 14:01). He says he trusts Meta because it is legible, profitable and suable (Sam · The Internet Is Fighting Back (Oct 2026) ▶ 14:25). In the same episode of June 2026 he relayed that hackers stole Instagram handles through Meta’s AI support for four months (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 24:54). That is his relay and unverified.

Routing and token spend

Sam’s line is that there is no one god model (Sam · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 24:40). His working stack in June 2026 was Gemini for sensitive finance and health, Grok for images and a central Claude setup, and he said he would switch to OpenAI in a day if Claude vanished (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 19:04). He also says he flips models on OpenRouter for the cheapest (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 19:37).

Dave describes the practitioner version. OpenClaw users, he says, have built skills that route to different models in real time to lower cost (Dave · OpenClaw vs Meta vs OpenAI: Th (Feb 2026) ▶ 52:28). He switched GLM 5.2 in for almost every utilitarian use in his agents (Jul 2026) (Dave · Open Source AI Wins. Now Labs (Jul 2026) ▶ 06:04). Sam, by contrast, says he does not care enough about cost to switch and will pay for slightly better accuracy (Sam · Open Source AI Wins. Now Labs (Jul 2026) ▶ 08:00).

On measurement, Sam says to judge spend by margin and profit, not volume (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 21:58). Brit tells companies to get staff to ask the agent first, then use tokens efficiently (Brit · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 31:23). Jess defines “token maxing” as ranking staff by tokens used, which also maximises cost (Jess · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 30:44).

The numbers are loose:

  • Sam guesses a few thousand dollars a month on his own bill (Sam · Chinese AI Model GLM 5.2 Beati (Jun 2026) ▶ 28:58), and admits he cannot measure the counterfactual (Sam · Google I/O Just Killed SEO? Ag (May 2026) ▶ 28:09).
  • Brit says her 11-year-old ran about $1,500 in a month (Brit · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 31:23).
  • Dave cites roughly 3,000 % more token use when agents reach an enterprise, with productivity per token unclear (Dave · Amir Efrati (May 2026) ▶ 20:39).
  • A speaker relays that an Uber CFO used a full annual token budget in 15 days, secondhand and hedged (“Uber CFO or something”) (one of the hosts · Elon Musk vs OpenAI Trial, Goo (May 2026) ▶ 19:44). Amir Efrati corrected a related Uber remark on air (Amir Efrati · Amir Efrati (May 2026) ▶ 19:30).
  • Brit found some recurring jobs spent far more than others and cut them back (Brit · Chinese AI Model GLM 5.2 Beati (Jun 2026) ▶ 29:38).

Sam argues the lack of measurable ROI per token is a weird moat for a while (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 21:58), and that CFOs take control once spend moves from experiment to core budget (Sam · Every CTO Is Quietly Switching (Jul 2026) ▶ 45:10).

Rent tokens or own the compute

Framework Rent tokens or own the compute covers this. Dave’s version is rent per token from Anthropic or OpenAI indefinitely, or pay up front for hardware, with a router as the middle path (Dave · Google I/O Just Killed SEO? Ag (May 2026) ▶ 08:04). His concrete reports:

  • A desktop Nvidia box at about $5,000 running a model of around 120 billion parameters, which he says runs OpenClaw well (Dave · Why the Met Gala Hates Tech, E (May 2026) ▶ 12:40).
  • GLM 5.2 on eight rented Nvidia B200s with friends (Jun 2026) (Dave · Chinese AI Model GLM 5.2 Beati (Jun 2026) ▶ 27:24).
  • A stock of B200 servers at about $500,000, for unlimited tokens at the cost of power (Dave · Every CTO Is Quietly Switching (Jul 2026) ▶ 43:39).
  • A hedge-fund friend’s report that Nvidia spot rental fell about 22 % in a week (Dave · Every CTO Is Quietly Switching (Jul 2026) ▶ 44:18).

Sam supplies the finance side: with financing and equal control, own it, unless frontier models are far better (Sam · Google I/O Just Killed SEO? Ag (May 2026) ▶ 09:32). He claims Anthropic tokens cost about 100 times open-source tokens (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 21:19). Dave concedes he had not tried local inference on a Mac Studio by March (Dave · Fire Sam Altman, The End of So (Mar 2026) ▶ 20:45).

At work

Most company examples come from the hosts’ own firms. Jess emailed her bot to build a headline-scoring tool and iterated on it in a meeting (Jess · Meta’s $17B Reckoning & Why Pa (Aug 2026) ▶ 05:47). Sam proposes a “master of bots”, the re-rising sysadmin (Sam · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 03:29), and mining recorded click-streams to find tasks to automate (Sam · Google’s AI-First Laptop, Meta (May 2026) ▶ 26:35). Dave coins “priority slop” for agents without context filling your inbox (Dave · ⁠The Real Reason Elon Wants Sp (Oct 2026) ▶ 36:22), and says email is now other people’s agents’ to-do list for your life (Dave · ⁠The Real Reason Elon Wants Sp (Oct 2026) ▶ 36:04). Guest Guillermo Rauch of Vercel says even good internal agents need human feedback (Guillermo Rauch · Guillermo Rauch (Apr 2026) ▶ 22:51).

How the view moved

  • Jan 2026. Sam calls local agents a security nightmare (Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 08:53) and predicts OpenClaw will be nothing in six months (Sam · Agentic AI Will Break the Inte (Jan 2026) ▶ 16:08). He ran it on a spare Mac Mini (Sam · Agentic AI Will Break the Inte (Jan 2026) ▶ 10:57).
  • Feb 2026. Dave announces the foundation is being structured (Dave · OpenClaw vs Meta vs OpenAI: Th (Feb 2026) ▶ 06:07); Sam says he runs the agent himself over Telegram (Sam · Elon’s Trillion-Dollar Move, W (Feb 2026) ▶ 33:23).
  • Mar 2026. Dave says the foundation is a Delaware nonprofit (Dave · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 01:46); Sam calls OpenClaw a thin shim (Sam · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 34:08).
  • Apr 2026. Sam moves to cloud instances and a bot-only email (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 21:32), and predicts a split between a fast, insecure crowd and a slower, costlier secure core (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 13:08).
  • May to Jul 2026. Dave’s talk shifts from routing to owning hardware (Dave · Google I/O Just Killed SEO? Ag (May 2026) ▶ 08:04) (Dave · Every CTO Is Quietly Switching (Jul 2026) ▶ 43:39).
  • Sep 2026. Dave says sandbox escape is an emergent property of training models to run longer (Dave · Suds (Sep 2026) ▶ 17:42).

Where the hosts disagree

  • How much access to give. Sam says the as-fast-as-possible crowd will win even though security suffers (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 12:31). Dave and Brit favour guardrails, trust splits and drafting without sending (Dave · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 20:32) (Brit · Suds (Sep 2026) ▶ 22:09).
  • Is OpenClaw durable? Sam: copyable, no switching cost (Sam · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 34:08). Dave: trust in a foundation-run community is the point, not lock-in (Dave · Nvidia’s GTC Was “Taylor Swift (Mar 2026) ▶ 35:06).
  • Can an agent replace an employee? Dave says no (Dave · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 17:10); Sam would keep Claude over his associate (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 17:50).
  • Are routers a business? Sam says an open-source router could be powerful (Sam · Every CTO Is Quietly Switching (Jul 2026) ▶ 47:09); Scott Belsky expects routers to be commoditised quickly (Scott Belsky · Peter Deng & Rachel Holt (Aug 2026) ▶ 25:36), and Peter Deng doubts they survive without a way to make a model your own (Peter Deng · Peter Deng & Rachel Holt (Aug 2026) ▶ 26:50).
  • Is local compute real? Dave calls it the wild card (Dave · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 37:38); an unattributed host would never run anything locally (one of the hosts · Meta’s AI Campaign, While Riva (Jul 2026) ▶ 37:05).

What could make this wrong

  • Dave’s figures on OpenClaw growth are rough and unsourced (Dave · Agentic AI Will Break the Inte (Jan 2026) ▶ 17:00).
  • Speed and price figures for hardware are self-reported (Dave · Every CTO Is Quietly Switching (Jul 2026) ▶ 43:44).
  • Sam’s 100x price claim is his own (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 21:19).
  • Sam says measuring open-source against frontier share is suspect because tokens are not comparable (Sam · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 27:32).
  • Sam’s crude risk analogy is partly a bid for clicks, by his own admission (Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 15:44).
  • Dave warns that an agent allowed to edit everything about itself can edit itself out of existence (Dave · Guillermo Rauch (Apr 2026) ▶ 24:55).

How strong is the evidence

Weak. The material comes from roughly 20 episodes between Jan and Oct 2026, but most is two or three hosts describing their own set-ups. Routing and token efficiency rest mainly on Sam and Dave. Sandbox habits rest on Sam, Dave, Brit and Jess in about four episodes, with no security specialist. OpenClaw’s adoption claims rest on Dave alone and carry his stake. The recurring question Is it safe to give an agent access to your accounts? gathers five answers from four episodes, three of them Sam’s. Will normal people use AI agents? shows that normal-person use is still early, with no usage data.

Go deeper

Sources

  1. (01)

    Google's AI-First Laptop, Meta's Spy Games, AI Monks in Middle America

    YouTube

    26:35 into “Google's AI-First Laptop, Meta's Spy Games, AI Monks in Middle America”

  2. (02)

    Is Google About to Kill Canva? │Claude Goes Rogue, Tesla vs. Waymo

    YouTube

    17:10 into “Is Google About to Kill Canva? │Claude Goes Rogue, Tesla vs. Waymo”

  3. (03)

    Why the Met Gala Hates Tech, Elon vs OpenAI Drama, and the Rise of Chinese AI Models

    YouTube

    12:40 into “Why the Met Gala Hates Tech, Elon vs OpenAI Drama, and the Rise of Chinese AI Models”

  4. (04)

    SaaS Companies Beware: AI Is The New UI (Anthropic's Claude Code and Cowork)

    YouTube

    6:33 into “SaaS Companies Beware: AI Is The New UI (Anthropic's Claude Code and Cowork)”

  5. (05)

    OpenClaw vs Meta vs OpenAI: The Personal Agent Wars Heat Up

    YouTube

    6:07 into “OpenClaw vs Meta vs OpenAI: The Personal Agent Wars Heat Up”

  6. (06)

    iPhone Duo, Meta Muse & Why AI Finally Scared Dave Morin

    YouTube

    22:35 into “iPhone Duo, Meta Muse & Why AI Finally Scared Dave Morin”

  7. (07)

    Meta’s $17B Reckoning & Why Parents Can’t Compete With Big Tech's Algorithm

    YouTube

    5:47 into “Meta’s $17B Reckoning & Why Parents Can’t Compete With Big Tech's Algorithm”

  8. (08)

    The Seed Market Is Broken: Mega-Rounds, Narrative Games, and Walkouts

    YouTube

    24:06 into “The Seed Market Is Broken: Mega-Rounds, Narrative Games, and Walkouts”

  9. (09)

    Open Source AI Wins. Now Labs Are Running to Washington |Thinking Machines, Johnny Ive, Realtime API

    YouTube

    6:04 into “Open Source AI Wins. Now Labs Are Running to Washington |Thinking Machines, Johnny Ive, Realtime API”

  10. (10)

    Elon Musk vs OpenAI Trial, Google Cloud Surge, Meta’s Blocked Acquisition, Anthropic Winning

    YouTube

    19:44 into “Elon Musk vs OpenAI Trial, Google Cloud Surge, Meta’s Blocked Acquisition, Anthropic Winning”

  11. (11)

    Allbirds Pivoted to AI Data Centers | SpaceX $1.5T IPO, Anthropic $800B, $175M Seed Rounds

    YouTube

    30:15 into “Allbirds Pivoted to AI Data Centers | SpaceX $1.5T IPO, Anthropic $800B, $175M Seed Rounds”

  12. (12)

    Apple, OpenAI & Why Nobody Makes Money in AI | The SaaS Apocalypse

    YouTube

    19:37 into “Apple, OpenAI & Why Nobody Makes Money in AI | The SaaS Apocalypse”

  13. (13)

    OpenAI's $852 Billion Valuation Makes Zero Sense (ClawCon Tokyo, Anthropic's Leak, Goodbye Internet)

    YouTube

    12:31 into “OpenAI's $852 Billion Valuation Makes Zero Sense (ClawCon Tokyo, Anthropic's Leak, Goodbye Internet)”

  14. (14)

    AI Can Change Everything and Still Be a Terrible Business + Peptide 101 from Matt Mazzeo

    YouTube

    40:28 into “AI Can Change Everything and Still Be a Terrible Business + Peptide 101 from Matt Mazzeo”

  15. (15)

    Why OpenAI Can't Win The Decentralized AI Future (OpenClaw, Apple's Win, X.AI Exodus)

    YouTube

    13:53 into “Why OpenAI Can't Win The Decentralized AI Future (OpenClaw, Apple's Win, X.AI Exodus)”

  16. (16)

    Agentic AI Will Break the Internet (ClawdBot, OpenAI & the AI Demand Delusion)

    YouTube

    8:31 into “Agentic AI Will Break the Internet (ClawdBot, OpenAI & the AI Demand Delusion)”

  17. (17)

    Normal People Don't Want AI Agents (And Silicon Valley Can't Accept It)

    YouTube

    19:30 into “Normal People Don't Want AI Agents (And Silicon Valley Can't Accept It)”

  18. (18)

    Meta's AI Campaign, While Rivals Ask Washington To Slow it Down | Lilian Weng Saga, China Cracks DUV

    YouTube

    11:58 into “Meta's AI Campaign, While Rivals Ask Washington To Slow it Down | Lilian Weng Saga, China Cracks DUV”

  19. (19)

    Nvidia’s GTC Was “Taylor Swift for Straight Men” (Apple, OpenAI, and OpenClaw News)

    YouTube

    1:46 into “Nvidia’s GTC Was “Taylor Swift for Straight Men” (Apple, OpenAI, and OpenClaw News)”

  20. (20)

    Fire Sam Altman, The End of Software Engineers, and Why AI Is All Narrative

    YouTube

    20:45 into “Fire Sam Altman, The End of Software Engineers, and Why AI Is All Narrative”

  21. (21)

    The Internet Is Fighting Back Against Your AI Agent

    YouTube

    13:20 into “The Internet Is Fighting Back Against Your AI Agent”

  22. (22)

    The Risk Of Securitizing GPUs Like Long-Lived Infrastructure & More - A MOL Special Team Edition

    YouTube

    25:36 into “The Risk Of Securitizing GPUs Like Long-Lived Infrastructure & More - A MOL Special Team Edition”

  23. (23)

    SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.

    YouTube

    24:40 into “SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.”

  24. (24)

    ⁠The Real Reason Elon Wants SpaceX and Tesla to Merge Now

    YouTube

    36:04 into “⁠The Real Reason Elon Wants SpaceX and Tesla to Merge Now”

  25. (25)

    Google I/O Just Killed SEO? Agentic Search, Universal Cart, and the New Web

    YouTube

    5:16 into “Google I/O Just Killed SEO? Agentic Search, Universal Cart, and the New Web”

  26. (26)

    Chinese AI Model GLM 5.2 Beating Frontier Models | Meta Glasses, Polymarket Scandal, AI Talent War

    YouTube

    27:24 into “Chinese AI Model GLM 5.2 Beating Frontier Models | Meta Glasses, Polymarket Scandal, AI Talent War”

  27. (27)

    Why SpaceX's $1.75T IPO Makes No Financial Sense (And Doesn't Have To)

    YouTube

    17:10 into “Why SpaceX's $1.75T IPO Makes No Financial Sense (And Doesn't Have To)”

  28. (28)

    Vercel CEO: 70% of Our Traffic Is Now AI Agents "Nobody Was Prepared" | Anthropic, OpenClaw, OpenAI

    YouTube

    22:51 into “Vercel CEO: 70% of Our Traffic Is Now AI Agents "Nobody Was Prepared" | Anthropic, OpenClaw, OpenAI”

  29. (29)

    Elon’s Trillion-Dollar Move, WaPo Layoffs & The Super Bowl

    YouTube

    1:26 into “Elon’s Trillion-Dollar Move, WaPo Layoffs & The Super Bowl”

  30. (30)

    Every CTO Is Quietly Switching Off Anthropic and OpenAI | GLM 5.2, Vox Americanus, Claude Fable 5

    YouTube

    43:39 into “Every CTO Is Quietly Switching Off Anthropic and OpenAI | GLM 5.2, Vox Americanus, Claude Fable 5”

(By email)

Compounding Insights