Navigate Select ESC Close

AIMM: A model for thinking about AI and Work

2025-11-20 Science & Technology
872
46
1
Unsupervised Learning
Unsupervised Learning
673.0k subscribers

Unlock all features

FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.

Description

This video introduces a 5-level AI Maturity Model to help you understand the rapid shift from generic chatbots to autonomous agents and business workflows. I demonstrate my open-source personal assistant, Kai, and showcase how to bootstrap your own personal AI infrastructure using free, open-source tools to magnify your productivity. 🎬 RELATED AI VIDEOS: • Building Your Own Unified AI Assistant Using Claude Code - https://youtu.be/iKwRWwabkEc • How My Projects Fit Together - https://www.youtube.com/watch?v=5x4s2d3YWak • The Future of Hacking is Context - https://www.youtube.com/watch?v=UwTTcka1Wd8 • Introducing Substrate - https://www.youtube.com/watch?v=ky7ejowc_qY • The 4 AAAAs of the AI ECOSYSTEM - https://www.youtube.com/watch?v=Pf6ydbCgU3U • Is OpenCode as Smart as Claude Code? - https://www.youtube.com/watch?v=yTylDxgyJZ8 • Claude Code + Neovim via Ghostty Panes - https://www.youtube.com/watch?v=ysVmQ6mesWE 🤖 OTHER TECH DISCUSSED: • Claude Code: https://claude.ai/code • Fabric AI Framework: https://github.com/danielmiessler/fabric • PAI: https://github.com/danielmiessler/Personal_AI_Infrastructure ✍🏼 RELATED BLOGS: • Building a Personal AI Infrastructure (PAI) - https://danielmiessler.com/blog/personal-ai-infrastructure • MCPs Are Just Other People's Prompts and APIs - https://danielmiessler.com/blog/mcps-are-just-other-peoples-prompts-and-apis • One-Click MCP Servers on Cloudflare - https://danielmiessler.com/blog/one-click-mcp-servers-cloudflare • Launching Daemon: Personal API - https://danielmiessler.com/blog/launching-daemon-personal-api • OpenCode vs Claude Code - https://danielmiessler.com/blog/opencode-vs-claude-code • Claude Code + Neovim via Ghostty Integration - https://danielmiessler.com/blog/claude-code-neovim-ghostty-integration • Web Scraping with BrightData + Claude Code - https://danielmiessler.com/blog/webscraping-with-brightdata-claude-code • Why Marcus Is Wrong About AI - https://danielmiessler.com/blog/why-marcus-is-wrong-about-ai • Why Dwarkesh Is Wrong About AGI - https://danielmiessler.com/blog/why-dwarkesh-is-wrong-about-agi • AI Is Becoming Like Reading - https://danielmiessler.com/blog/ai-becoming-reading • We Have Enough AI for AGI - https://danielmiessler.com/blog/we-have-enough-ai-for-agi • The Worst AI Metric - https://danielmiessler.com/blog/the-worst-ai-metric • AI Workforce Volume-Difficulty Curve - https://danielmiessler.com/blog/ai-workforce-volume-difficulty-curve • Survive AI: Become Creators - https://danielmiessler.com/blog/survive-ai-become-creators • The Great Bifurcation - https://danielmiessler.com/blog/great-bifurcation • I'm Worried It Might Get Bad - https://danielmiessler.com/blog/im-worried-it-might-get-bad • AI Model Ecosystem: 4 Components - https://danielmiessler.com/blog/ai-model-ecosystem-4-components • Using the Smartest AI to Rate Other AI - https://danielmiessler.com/blog/using-the-smartest-ai-to-rate-other-ai 🔗 CONNECT: X: @danielmiessler Website: https://danielmiessler.com GitHub: https://github.com/danielmiessler/ I'm building sick shit over here, and I've only just begun. Let's build together. Subscribe to the channel here and catch the newsletter below. See you in the next one! 📧 NEWSLETTER: https://danielmiessler.com/subscribe

Top Comments (7)

@liamvnvliet 2025-12-23

I’m curious, you put workflows at 3, but I see workflows as a chain of sorta set actions while the agent is much more self sufficient long running and adaptable So my question is: why do you see workflows as 3 and not 2 I get the documenting of it, but much of those blue parts is made up of lots of choices and different kinds of actions that depend on the context. Humans are good at undertstanding the context, talking to other humans and then imputing missing information and remembering information. I rather want an agent than a pipeline. That matches the strong parts of a human. The hurdle I see is currently talking to humans naturally, getting context and remembering information for a long time and working continuously for months. If that gets solved for devs to build easily, then we get into solving knowledge jobs and their complex workflows.

1
@aojsr9986 2025-12-10

I can't seem to get the voices to run consistently like you do. Love your videos btw

0
@MitchemBoles 2025-11-20

Fascinating walk through and take on the escalating agentic capability and usage by everyone, but more specifically cyber attackers and defenders stages at the end. Great model to think through. No doubt AI Workflows at Level 3 are already in use, but Anthropic's delineation shows that Level 2 attackers are ahead of the curve with commodity AI tools and a few bucks. Path of least resistance for attackers will be the easiest choice, and this only shows us the need is very real for defenders to implement strategies and tools that actually work to defend.

0
@gavinslater5117 2025-11-22

Loving you videos Daniel...another very insightful one!

0
@RF-js9yj 2025-12-16

This is a custom GUI using Grafana? How to set up my own?

0
@cypsteel 2025-11-22

I really appreciate your insight and knowledge. I’ve been using Claude Code for about three weeks now, and the file permanence plus local interactivity has completely changed how I use AI. You mentioned getting everyone at work on agentic AI “like Kai.” What does that look like in your mind? For example, if someone is on a Security team, what would you have an agent like that doing?

0
@emotionalmindedstate 2025-12-20

5 AI shell. AI surrounds the human and becomes human extension.

0

Unlock the Data Inside
Turn Videos into Knowledge

  • Get FREE 10/day: transcripts, summaries, chats
  • Chat with videos, export text & PDF
  • $1 free API credit for RAG, chatbots & research

Free forever plan • All features unlocked

App screenshot