dumbfoundry

@dumbfoundry · dumb is fun

Deep research on sovereign AI, private compute and the agent economy, turned into Shorts, Trailers and full Videos. Every video has a free research report here on the site.

205subscribers
135videos
108Shorts
156reports
347topics

Latest episode

Agent Zero Lets You Approve Each Action

Agent Zero 2.15, released 1 November 2026, changes how you decide what your AI assistant is allowed to do. Instead of one yes-or-no switch per tool, you can now approve or block individual actions inside each tool: read my files, yes; rewrite them, no. Every action gets three positions — On, Default or Off — so new tools follow your house rule automatically instead of arriving pre-approved. The release also closes two quiet loopholes. Telegram and WhatsApp senders can no longer lift their own restrictions; settings changes now require explicit administrators. And the remote typing tool now needs the same permission as the remote terminal, so blocking one really blocks both. Think of it as moving from handing over the whole key ring to cutting keys room by room. The same question applies to every app in your life: what can it do without asking you first, and who else can tell it what to do? #AgentZero #AIAgents #LeastPrivilege #AISecurity #OWASP #Privacy #OpenSourceAI — Links — https://agent-zero.ai/ https://github.com/agent0ai/agent-zero https://genai.owasp.org/llmrisk/llm062025-excessive-agency/ https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/ https://csrc.nist.gov/glossary/term/least_privilege a dumb drop by dumbfoundry

📄 Read the report → ▶ Open on YouTube ↗

Recent videos

Who Gets the Keys? House Rules for Your AI

2026-11-02 demo

#agent zero #https #release notes #the whole #ai-savvy builders #allowed #business and non-technical #custom integrations #dumb drop #excessive agency #five steps #give

Satellites Show Which AI Data Centers Are Real

Companies announce AI data centers. Satellites show whether they are real. This video breaks down a business built on watching the AI buildout from space, then turning those images into market signals that buyers actually pay for. We look at how commercial and free satellite imagery can track construction progress, cooling infrastructure, power connections, and thermal signatures at data centers, chip fabs, and energy sites. The key product is not raw imagery. It is time-stamped intelligence: capacity estimates, go-live detection, timeline deviations, and risk flags compared with public guidance. Then we get into the operating model. Agent Zero agents run continuous monitoring, change detection, object counting, data fusion, and report generation. A0T recurring compute and subsidized inference pathways keep the agent swarm cheap enough to run all day. Adaptive escalation keeps expensive high-resolution and thermal imagery focused only on sites where something changed. Key topics: - Why satellite receipts beat corporate announcements - Cooling units and thermal data as power and compute proxies - How Agent Zero automates collection, analysis, and delivery - A0T compute economics for always-on agent swarms - Imagery budget tiers from starter to institutional - Selling derived signals through APIs, alerts, dashboards, and subscriptions - Margins, error bars, and the risks in power estimates If you want more breakdowns of AI infrastructure, alternative data, and autonomous agent businesses, like, subscribe, and tell us in the comments which site you would monitor first. #AIInfrastructure #SatelliteImagery #AgentZero #A0T #AlternativeData #DataCenters #AutonomousAgents #Compute — Links — https://agent-zero.ai/ https://github.com/agent0ai/agent-zero a dumb drop by dumbfoundry

2026-11-01 demo

#agent zero #satellite imagery #ai data centers #ai infrastructure #alternative data #thermal imaging #compute economics #market signals

Satellites and Agent Zero Track AI Data Center Buildout

2026-11-01 demo

The AI Boom Is Now Funded by Debt — And Your Pension Is the Lender

The AI boom quietly changed hands. It used to be paid for by shareholders buying a story about the future. Now it's paid for by lenders who get a fixed coupon and don't care how magnificent the company becomes. That single shift changes every question worth asking. Goldman Sachs senior multi-asset investor Alexandra Wilson-Elizondo argues that "is it a bubble" is now the wrong question — not because the answer is no, but because it's a shareholder's question, and the important money isn't shareholders' money anymore. In credit, you don't underwrite a narrative. You underwrite cash flow, credit quality and counterparties. This video breaks down the five sharper questions that replace the bubble debate, why a sold-out sign on AI compute through 2027 isn't the same as profitable factory time, and the mechanical reason lenders are getting nervous: credit market indigestion. When a queue of borrowers each needs billions, the middle of the queue pays more. A firm that planned to borrow at 5% and now faces 7% pays £2 million more a year on a £100 million loan — for the exact same plan. Then the part almost nobody mentions: who is actually lending this money? Pension funds. Insurers. Bond funds holding ordinary people's savings. You may already own a slice without ever choosing it. Key takeaways: - Why debt funding moves the decision from fans to lenders - The five questions that replace "is it a bubble" - Credit indigestion and the arithmetic of a higher coupon - Why the weakest borrowers, not the hyperscalers, carry the risk - The 4% safe-rate hurdle every risky bet has to beat - How ordinary savers end up as the counterparty If this reframed how you think about the AI trade, like, subscribe, and tell us in the comments: do you know what your fund actually holds? #AIBoom #CreditMarkets #AIInvesting #DebtFinancing #GoldmanSachs #PensionFunds #MarketAnalysis #AIEconomy — Links — https://agent-zero.ai/ https://github.com/agent0ai/agent-zero https://www.ft.com https://www.goldmansachs.com https://www.wsj.com https://www.factset.com https://www.spglobal.com/marketintelligence https://www.goldmansachs.com/ https://www.ft.com/ a dumb drop by dumbfoundry

2026-10-31 analysis

#ai-boom #credit-markets #debt-financing #ai-investing #pension-funds #market-analysis #goldman-sachs

Inside the AI Boom's Real Lender: Credit Quality, Coupons and the 4% Hurdle

2026-10-31 analysis

#ai-boom #credit-markets #debt-financing #ai-investing #pension-funds #market-analysis #goldman-sachs

64 Cents for a Dollar of AI Compute — And the Catch Nobody Mentions

One token costs $1.38 and sits on top of roughly $2.16 of AI compute. That gap is either the deal of the year or a warning label — and this video walks the arithmetic without flinching either way. Agent Zero Token (A0T) is a staking token that pays you back in computer time instead of dollars. Lock it and you get a Venice API key that refreshes daily, giving you access to AI models that don't attach your name to the request. Lock longer and your share of the pool multiplies up to fifteen times, with a daily pool multiplier stacking another two to three times on top. Run the numbers and the payback swings from 575 days with no lock to about 31 days with a five-year lock. But that fast payback isn't a gift. It's the price of your tokens being unsellable for half a decade. And the payback is measured in usage — if you never burn the credit, the clock never starts. Then there's the uncomfortable part: daily trading volume of roughly $3,700 against a $1.38 million paper valuation. A puddle, not a pool. The quoted market cap is the last trade times a million; nobody ever put that money on the table. A token-security scanner also flags that the contract creator can still change fees, disable sells, or mint more tokens. In this video we break down the valuation gap, the staking multiplier math, the liquidity problem, and the honest question every viewer should ask first: what's your actual monthly AI bill? Key takeaways: • Why A0T trades at 64 cents per dollar of its own reserve • How staking multipliers turn 575 days into 31 days • Why thin liquidity makes the "market cap" largely fictional • The contract-level risks a security scanner already flagged • Who this is genuinely for — and who should just close the tab If you want more breakdowns that read the fine print instead of the pitch, like, subscribe, and tell us in the comments whether a 36% discount is enough to compensate for a five-year lock. #AgentZeroToken #A0T #VeniceAI #CryptoAnalysis #AICompute #Tokenomics #DeFi #CryptoRisk — Links — https://agent-zero.ai/ https://github.com/agent0ai/agent-zero https://www.coingecko.com/en/coins/agent-zero-token https://basescan.org/token/0xCc4ADB618253ED0d4d8A188fB901d70C54735e03 https://www.geckoterminal.com/base/tokens/0xcc4adb618253ed0d4d8a188fb901d70c54735e03 https://gopluslabs.io/token-security/8453/0xCc4ADB618253ED0d4d8A188fB901d70C54735e03 https://www.mexc.com/price/agent-zero-token https://venice.ai/ https://www.coingecko.com/ a dumb drop by dumbfoundry

2026-10-30 analysis

#agent-zero-token #a0t #venice-ai #tokenomics #crypto-staking #ai-compute #liquidity-risk #valuation-gap