Brutalist

Videos

Video tutorials, talks, and educational content.

Brutalist.art - The Slide Deck Is a Program

Brutalist.art - The Slide Deck Is a Program You've used Beautiful.ai. You've used Gamma. You walked away with a conclusion about what AI slides can do. That conclusion is wrong. And the gap between what you think is possible and what's actually possible is the entire point of this video. A browser doesn't display slides. It runs programs. It plays audio, draws live charts, adapts to any screen, responds to keyboard input — natively, without plugins, without export steps. That medium has existed for twenty years. It was locked behind a skill ceiling that kept everyone who wasn't a frontend developer out. Then the skill ceiling dropped. With Claude, you describe what you need in English. The code gets written. A living, deployable HTML deck — audio on every slide, animations with physical weight, responsive on any device, shareable as a URL — in the time it takes to have a conversation. That's not a smarter template. That's a different medium entirely. In this deck: What a canvas-based tool actually is (and what it can't do by design) Why the cargo cult version of AI slides never questions the canvas model The chef/menu analogy — and why it permanently replaces the template frame The Solari board moment: what happens when someone describes what they need instead of picking what's available Three steps to start: copy, paste, type help Get the Brutalist system prompt: → [link in pinned comment] See the deck this video was built with: → https://www.brutalist.art/talks/brutalist/brutalist-intro.html Explore all Brutalist talks: → https://www.brutalist.art/ Brutalist is part of the Humanitarians AI Ecosystem. Built by Nik Bear Brown · Bear Brown LLC Professor Bear · Teaching the thing underneath the thing. TIMESTAMPS 0:00 — You've made slides before. This isn't that. 0:30 — What you'll be able to do by the last slide 1:00 — The canvas model: what you're actually looking at in Beautiful.ai 2:00 — A browser is a runtime, not a display 2:45 — What Brutalist decks actually do: audio, animation, live data, one file 3:30 — Pause & reflect 3:45 — The cargo cult version: describe → template → export 4:30 — A marketing final, a URL, a voice on every slide 5:15 — Two ways to get a meal 5:45 — The chef conversation 6:30 — Pause & reflect 6:45 — Stop asking what's available. Start asking what you need it to do. 7:15 — The Solari board moment 7:45 — Three steps: copy, paste, type help 8:00 — Close #BrutalistDesign #AISlides #ClaudeAI #HTMLPresentation #ProfessorBear #InstructionalDesign #NoCode #AITools #PresentationDesign #BackwardsDesign

Why Every Claude Code Prompt You've Written Has the Same Structural Failure

Most Claude Code users are playing an instrument. Conductors build production sites in three hours. Here's the difference — and why it starts before you write a single prompt. This video breaks down the solve-verify asymmetry: why Claude produces confident, fluent, structurally wrong output when the problem is poorly framed, and what that means for anyone using AI to build anything real. On March 30th, 2026, Boondoggling.ai went from idea to live in roughly three hours — six routes, a hybrid file system and database architecture, an admin dashboard, a community upload pipeline, and a full prompt library. Not a toy project. A real build. Twenty steps, nine Claude tasks, eleven human tasks, one hour with Claude Code. Every prompt ran without error. The reason isn't a better prompt. It's the five supervisory capacities: plausibility auditing, problem formulation, tool orchestration, interpretive judgment, and executive integration. These are the conductor's tools. They're the subject of the Conducting AI course in the Irreducibly Human series. And they're what separates a build that ships from a session that spirals. The Gru system prompt — the senior software architect persona used in this build — is free at boondoggling.ai/tools. No account. No API key. No paywall. Paste it into a Claude project and type /help. 🔗 Try Boondoggling.ai → https://boondoggling.ai 🔗 Irreducibly Human series → https://irreducibly.xyz 🔗 Gru system prompt → https://boondoggling.ai/tools TIMESTAMPS 0:00 — The conductor vs. the player 0:30 — Why Claude prompts fail structurally 1:15 — The solve-verify asymmetry 2:00 — The five supervisory capacities 4:30 — The Boondoggling.ai live build (March 30, 2026) 6:00 — The boondoggle score and handoff conditions 7:30 — How to run Gru in your own Claude environment TAGS: Claude Code tutorial, AI prompting mistakes, how to use Claude Code, supervisory capacities AI, conducting AI, Boondoggling AI, Grue system prompt, Claude project instructions, AI build methodology, problem formulation AI, plausibility auditing, tool orchestration AI, irreducibly human series, Professor Bear, Nik Bear Brown, AI conductor, Claude AI workflow, AI code generation, human AI collaboration, AI prompting framework HASHTAGS: #ClaudeCode #AIPrompting #ProfessorBear

Nik Bear BrownNikBearBrown.com

Boondoggling: You Are the Conductor

What Most Developers Miss About AI-Assisted Programming https://www.skepticism.ai/p/boondoggling-you-are-the-conductor?utm_source=youtube

Nik Bear BrownNikBearBrown.com

What Is Medhavy?

What Is Medhavy? Medhavy is an adaptive learning platform built around one question: does this actually help the learner? Not a feature set. Not a product category. A frame — and a continuously running experiment on what AI-powered learning should actually look like. Most platforms pick a model and defend it. Medhavy runs the test: on each learner, each teaching approach, and each AI system as the technology evolves. The results stay visible. That's not a feature. That's the condition under which trust becomes possible. The platform operates on a strict priority order: learner first, instructor second, organization third. If learners are only there because they're required to be, the platform has failed by its own definition. Instructors build learning experiences that reflect their actual pedagogical vision — not a compromise with a rigid template. Institutional constraints like compliance, branding, and accreditation are real, and they're easy to configure — but they're the last layer, not the first. The underlying AI systems are specialists. Each one is shaped by an instructor's intent, each one in service of a learner who chose to be there. What's producing genuine learning outcomes? What just feels impressive? The experiment runs on both questions at once. This video breaks down the full Medhavy frame — what drives every product decision, how the three-layer priority works in practice, and why the experiment is the product. → Learn more at medhavy.ai Medhavy LLC · in collaboration with Bear Brown LLC · Humanitarians AI

Nik Bear BrownNikBearBrown.com

Irreducibly Human: What AI Can and Can't Do

AI can outrun you on facts, patterns, and arithmetic. What it can't replicate is human judgment — and that's the only thing that matters now. Irreducibly Human is a curriculum series built around one question: what does AI actually need you to do? Not compete with it. Not fear it. Supply the reasoning it genuinely cannot supply on its own. This video introduces the series — seven tiers of human intelligence where either humans lead or nothing gets decided at all: causal reasoning, ethical judgment, collective intelligence, and the wisdom that only comes from having real stakes. No algorithm commits. No algorithm can lose. That gap is the curriculum. The series currently includes six courses and two companion books, each targeting a different place where human judgment is either irreplaceable or dangerously underdeveloped. It was built using the same AI tools it teaches you to oversee. That's not irony — that's the argument made concrete. Built for educators, curriculum designers, and anyone who creates learning for other people. The courses live at irreducibly.xyz. The curriculum is a working document — shaped by debate, sharpened through conversation. That's not a disclaimer. That's the design. 🔗 Explore the full series → https://irreducibly.xyz Chapters 0:00 — Welcome to Irreducibly Human 0:18 — The problem with how schools taught thinking 0:55 — What the series is and why it exists 1:45 — Who this was built for 2:05 — Subscribe TAGS: irreducibly human, AI and education, human judgment AI, what AI cannot do, future of learning, curriculum design AI, causal reasoning education, ethical judgment teaching, AI literacy educators, human skills artificial intelligence, education technology 2025, learning design AI, AI curriculum series, irreducible human intelligence, educator professional development HASHTAGS: #IrreduciblyHuman #AIandEducation #FutureOfLearning

Nik Bear BrownNikBearBrown.com

Sacred Emily (1913) — Gertrude Stein | Spoken Word by Nik Bear Brown

Sacred Emily (1913) — Gertrude Stein | Spoken Word by Nik Bear Brown This is a verbatim spoken-word reading from Sacred Emily (1913) by Gertrude Stein, as interpreted by Nik Bear Brown. Written during Stein’s most radical modernist period, Sacred Emily abandons conventional narrative and grammar in favor of repetition, rhythm, and sound, treating language itself as the subject. Best known for introducing the line “Rose is a rose is a rose is a rose,” the poem was influenced by Cubism and early experiments in perception, and is meant to be experienced aloud rather than interpreted linearly. This reading preserves Stein’s original text exactly, allowing the cadence, insistence, and musicality of the words to reveal their meaning through listening. Released via Musinique. #SacredEmily #GertrudeStein #SpokenWord #ModernistPoetry #SoundPoetry #ExperimentalLiterature #PoetryReading #NikBearBrown #Musinique #PublicDomainPoetry

MusiniqueMusicSpotifyApple MusicIndie MusicMusic AIMusic Algorithmsindie musiciansAI songwritingmusic analyticsstreaming fraudplaylist intelligenceprotest musicindependent artistsmusic technologyAI music toolsMusinique.com
Page 1 of 6