A 17-minute tier-list of one practitioner's actual daily stack — and the five mental models that keep him from drowning in new releases.
Staying ahead in AI isn't about adopting every new tool — it's about keeping a lean, tiered stack and applying a North Star test to filter every new release before it pulls you off your path.
The video argues that AI overwhelm is a focus problem, not an information problem. The presenter's actual daily stack is intentionally thin: Claude Code as primary OS, VS Code as the IDE host, Glaido for speech-to-text. Everything else is tiered by frequency of actual use rather than hype. The second half introduces five mental models — the North Star Test, the 20% Productivity Dip Rule, needle-moved-per-hour productivity, tool-agnostic directory architecture, and a decision flowchart — that together form a system for evaluating new tools without getting pulled off the path.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →
Frames the problem: too many tools, too fast a pace. Promises a real daily-use tier list plus the mindset to filter new releases.
Claude Code ("basically my operating system"), VS Code as the IDE host, Glaido as speech-to-text. These three cover his entire day.
Codex (complements Claude Code's weaknesses), Claude Chat (quick one-offs), Hermes Agent (Telegram-triggered, out-and-about work), Perplexity (agent research), Grok/X (Twitter thread mining).
Tools used for one specific job inside a process: Apify for scraping, GPT Image 2 for creation, Nano Banana 2 for editing, key.ai for image/video model routing, HeyGen for avatars, ElevenLabs for voice cloning, OpenRouter for model switching.
Gemini and Windsurf rarely touched. Ollama for open-source model tracking. Manus for occasional tests — noted as a valid S-tier for complete beginners.
ChatGPT, OpenClaw, Cursor, NotebookLM, Poppy AI, n8n, WhisperFlow — retired not because they're bad but because their value was absorbed into more central tools or custom-built into the stack.
Promo for free Skool community resource guide containing all tools and frameworks from the video.
All agents run in the same project directory. Switch agents without rebuilding your project. Build directories to outlive any tool.
Define your actual outcome first. Every tool/feature that doesn't measurably advance it is a gravity well pulling you off course.
Every switch costs efficiency. Only worth it when the new ceiling is genuinely higher — not just a return to baseline.
Productivity is output per hour, not time logged. Stop collecting tools, start shipping. The right question is which tool is best for this specific sub-task, not which tool is globally best.
Flowchart for every new release: does it solve a real pain point now? No: save the link. Yes: 1-week real-scenario trial, then integrate or reject without guilt.
The antidote to AI overwhelm isn't more information about tools — it's a tiered mental model for when to adopt, when to wait, and when to retire.
“I could use just my Claude Code and my Glaido and just work the whole day and be fine.”
“Build your directories like they're going to outlive any tool. Because they will.”
“Productivity is needle-moved per hour, not hours worked.”
“The right question isn't which tool is best. The right question is: for this specific task in this specific context, which tool is best?”
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
Every week there's a new agent framework, a new model, a new must-have tool — and the guilt of falling behind compounds faster than anyone can ship. This breakdown cuts through the noise: here's the actual lean stack one AI builder uses every day, the tools he's quietly retired, and the five decision rules that keep him on the path when the feed is screaming at him to pivot.
Five-tier classification of every AI tool in the presenter's ecosystem by actual frequency of use.
All agent frameworks (Claude Code, Codex, Hermes, OpenClaw) are wrappers around underlying models that operate in a shared project directory. Portability is built in if the directory is built to be tool-agnostic.
Before adopting any tool: name the measurable outcome you're building toward, then ask if this tool/feature gets you measurably closer. If you can't articulate how, skip it.
Every workflow change causes a temporary ~20% efficiency dip. The change is only justified if the new performance ceiling after recovery is higher than the old one — not merely a return to baseline.
Hours worked is a vanity metric. Productivity is measured by actual outcomes per hour — not time logged. Twelve hours of reading and planning can be less productive than four hours of direct execution.
A seven-step flowchart for evaluating every new AI tool or feature drop without spending a month rabbit-holing on things that don't advance your current goal.
“I broke all of this down into a free resource guide — all these tools, all these different frameworks and rules to remember — and I put that inside of my free school community. The link for that is down in the description.”
Mid-video insert with direct community plug and classroom navigation walkthrough. Natural and contextually timed before the mindset section.
00:00
00:19
00:32
00:45
00:58
01:09
01:2301:36
01:49
02:02
02:15
02:28
02:41
02:54
03:07
03:20
03:33
03:45
03:5804:11
04:24
04:37
04:50
05:03
05:16
05:29
05:42
05:55
06:08
06:20
06:33
06:46
06:59
07:1207:25
07:38
07:51
08:0508:1508:28
08:42
08:5509:08
09:2109:36
09:47
10:00
10:13
10:2610:39
10:52
11:04
11:17
11:3011:43
11:56
12:09
12:22
12:37
12:48
13:01
13:14
13:27
13:39
13:52
14:05
14:18
14:28
14:4414:57
15:10
15:23
15:36
15:49
16:01
16:08
16:27
16:40
16:53
17:06
05:41Claude Code ships auto mode — a classifier-backed middle path between constant permission prompts and the anything-goes risk of dangerously skip permissions.
March 24th
09:43A 9-minute demo showing how native cron sessions turn any Claude Code skill into a self-healing, self-improving background agent.
March 7th
07:24A 7-minute case for front-loading knowledge extraction before you write a single line of your AI operating system.
June 4th
28:57A 29-minute walkthrough of the Four Cs framework for running your entire business through Claude Code.
May 29th
27:13A 27-minute live build that wires Claude Code into a context-aware AI that plans your day, runs research, and checks on your team, all in parallel.
March 5th
27:18A 27-minute walkthrough of Claude Code skills — what they are, how they stay cheap, and a live build that turns a Kie.ai API call into a reusable team automation.
February 27th