2026-10-06
In 'The oil apocalypse is here' (Max Fisher, YouTube, an October 2026 snapshot), Fisher argues that the gas-price spike is a distraction from a three-part crisis already underway. The first part is a crude supply shock from converging wars: Iran/Hormuz, Russia-Ukraine and Saudi Arabia-Houthis. The Houthis closed Bab al-Mandab and Iraqi militia drones destroyed the Saudi East-West pipeline's pumping stations. Fisher says Gulf output has fallen from about 20M to 8M barrels a day and the global net shortfall is about 7M b/d. Fisher reads China's behavior as a warning sign: it pays about $135 a barrel against about $105 elsewhere, and cuts imports by 2.5M b/d instead of 5.5M. The second part is a shipping shock. Rerouting around the Cape of Good Hope stretches voyages from 19 to 48 days, which works like losing two of every three tankers. Supertanker rates have gone from about $30,000 to $1.2M a day, which adds about $23 a barrel, roughly the size of the Iran-war price shock itself. The third part is a refining shock. Ukrainian drone strikes on Russian refineries, an Iran vs. Gulf-states refinery war and Chinese export cuts have removed about 3M b/d (around 10%) of global diesel. Fisher presents diesel as the channel through which all of this reaches everyday life. It is about 30% of the barrel, its demand is inelastic, and it is hit twice by freight (crude to the refinery, then product to the market). A record crack spread turns a roughly 60% crude rise into about a 250% diesel rise. Fisher traces that into trucking (about 3x), groceries (about +20% before farm costs), construction and housing, airfares (about +40%, with insolvencies and cancellations) and overseas trade (shipping about 3x). Fisher says a US diesel export ban would do 'worse than nothing' and that stocking up on fuel probably doesn't matter. Fisher's advice is not to treat this as the new normal, because the crisis grows with every week the war that triggered it drags on. All figures are Fisher's own estimates and will date quickly.
2026-09-29
In "How AI Is Changing the Way Product Managers Operate: From Information Mover to Builder with Judgment" (https://www.youtube.com/watch?v=yUohoaC8_Hs), Singhal argues that AI is upending product management: the 'information mover' PM, who spent the job relaying information between layers of management, is becoming obsolete, while builders with judgment are in high demand. He expects the mechanical parts of building to be automated within about two years, producing 10–100x more changes to products, so the PM's job shifts from moving information to evaluating what is good, worth shipping, sustainable and differentiated. He cites record pay and more offers for top builders, predicts a wave of staff shedding followed by AI-first rehiring over 12–24 months, and says non-builders (roughly half of product people) may have to leave tech or build outside it. Reviews, stand-ups and status reports will be automated with agents that supply ground truth, so alignment loses its theatrics but someone still has to decide what to fight for. Scale now comes from your own opinion and what you build, and a builder-PM should build tools to stay on top of the team's work rather than be the '51st engineer'. Non-engineers can start by being opinionated and telling a coding AI (Claude Code, Codex, Lovable) in plain English what they want. His habits include automating away your own tasks daily, colour-coding a week of meetings green, yellow and red, running 'show and tell' demos of AI builds, and standardising on a single AI tool. He says the main obstacle is psychological: those who mastered the old system find reinvention hardest, and a first-joy moment building something small and personal turns fear into energy and creates time. His thrive checklist for the next two years is to find joy, obsolete yourself, raise your pace, make time, swallow your ego about past titles and stay long-term focused. Hiring is also shifting toward modern delivery and scenario-based interviews over résumé brands. He also warns that geographic, age, gender and ethnic diversity may take a step back because the AI wave is Bay Area-centric, and that design roles are plateauing because the industry conflates 'pixel generators' with 'taste makers'. His optimism (a renaissance, 'smiling exhaustion') sits alongside his warning that about half of product people are in trouble.
2026-09-08
Fireship's video '5 open source tools that replaced my $320/mo AI stack...' argues that stacking paid AI subscriptions — Cursor, Claude Max, GPT Pro, Gemini Ultra, ElevenLabs, DeepSeek, plus forgotten API keys, totaling roughly $320/month — broke the subscription model (played for laughs via a jab that 'Big AI' is displacing US alcohol sales), so he canceled everything and rebuilt an equivalent stack self-hosted on a single Hostinger VPS: Ollama for local open-weight inference, Nine Router as an OpenAI-compatible proxy that waterfalls usage across paid, pay-per-token, and free tiers, Headroom to strip low-value tool/log output into a locally-cached, reversible compression layer before it hits billable tokens, Dify as a visual node-based app builder (demoed via a running 'Horse Tinder' matchmaking example), and Open Hands as an autonomous SWE-bench-Verified-ranked coding agent to resolve GitHub issues unattended. The stack isn't pitched as local-only — Nine Router deliberately preserves a fallback path into hosted Claude/GPT so frontier access survives even after local capacity or a declined card — and the piece closes on the more provocative (and joke-inflected) claim that Open Hands makes manual app-building 'outdated.' It's a fast, promotional, joke-dense tour (Hostinger is the sponsor) that names what each tool does and why to bolt it on, without going deep on configuration for any of them.
2026-09-08
"5 open source tools that replaced my $320/mo AI stack..." (https://www.youtube.com/watch?v=Y5rSSvXfL4g) is a sponsored, comedic tour arguing that Ollama, Nine Router, Headroom, Dify, and Open Hands — deployed together on a single VPS (Hostinger, the sponsor, via its one-click Docker catalog) — can replace a roughly $320/month bundle of paid AI subscriptions (Cursor $20, Claude Max $100, GPT Pro $100, Gemini Ultra $100, plus assorted extras) while staying cheaper and still leaving the option to tap Claude or GPT when actually needed. Ollama is pitched as 'Docker for LLMs,' running open-weight models locally with private prompts and zero inference cost (though frontier performance still needs frontier hardware); Nine Router sits in front of dozens of providers behind one local OpenAI-compatible endpoint and rolls requests through tiers — an existing paid subscription first, then a cheap pay-per-token model, then free providers — while also tracking usage and compressing tool output; Headroom is a context-compression layer that trims tool outputs and log files before they become billable input tokens, caching the originals locally so nothing is permanently lost; Dify is a visual, node-based workflow builder for assembling AI apps (demoed with a tongue-in-cheek 'Horse Tinder' matchmaking workflow exposed as an API) and Open Hands is a self-hostable autonomous coding agent, a top SWE-bench Verified performer, framed as Dify's open-ended complement for fixing GitHub issues on its own. The video stays functional rather than deep throughout, and its cost math and specific tool lineup are current-as-of-filming rather than durable claims.
2026-09-08
A 2026 YC Paper Club talk, 'Why The Harness Matters More Than The Model' (filed 2026-09-08), argues that the tool/memory/control-loop scaffolding wrapped around a frozen LLM — the 'harness' — now drives real-world agent performance as much as the underlying model, citing the same Claude Opus model jumping from 30% to 95% on the ARC-AGI private holdout purely by swapping harnesses with no model change, plus an 18-point swing between two harness versions on another eval; it traces six years of harness lineage from GPT-2's bare sampling loop through few-shot prompting, chain-of-thought, tool-calling (WebGPT, Toolformer), memory (MemGPT), skills (Voyager), code-as-action (Intercode), and self-reflection loops (ReAct, Self-Refine, Reflexion) up to today's static 'harness v1' (orchestrator plus tools, skills, sub-agents, and recursion), and claims the frontier has shifted over roughly the last six months toward self-improving harnesses that rewrite their own system prompt (DSPy-style search) or their own code (Darwin/Gödel-machine meta-harnesses). Concrete case studies include Prime Agent — an RLM/REPL-based 'continual harness' whose ARC-AGI runs swung wildly between 78%, 95.5%, and an invalidated 99.9% depending on model, while a rival harness reportedly burned about $5,000 for little gain — Open Jarvis's push to move personal AI on-device by having a cloud LLM optimize a local model stack offline for roughly 800x lower cost, and YC's own internal agent QM, which evolved from single shared agents to a design centralizing agent state in Postgres with disposable sandboxes, a deliberately thin three-tool core, and hard-won lessons like forcing a minimum 'grind' time/token budget before an agent is allowed to quit a task.