Active Deep depth Life-changing outcome

AI workflow lab

Less about hype and more about turning AI into an actual workbench for writing, synthesis, product architecture, and low-cost leverage.

Technical category AI Systems & Agents subcategory 3 cycles $190 spent ~180 hours in the current cycle time
research-heavy tag building tag learning tag teaching tag deep-dive tag monetizable tag online tag

What triggered it

Watching model quality cross the line from novelty to useful collaborator.

Goal

Build repeatable workflows that save time without creating a brittle dependency stack.

What I did

Tested models, built prompt templates, compared notes across tasks, and started designing products around what can actually be automated cheaply.

Summary notes

This one sticks because it produces immediate outputs. The dangerous part is mistaking tool exploration for product progress.

Lessons learned

  • The useful frontier is task design, not prompt theatrics.
  • Keeping outputs in structured notes makes the learning compound.

Mistakes and dead ends

  • Spent time comparing models before defining the actual workflow.
  • Saved too many prompts without tagging what problem they solved.

Related focuses

Same neighborhood

Homelab and hosting rebuild

A corrective cycle after overcomplicating hosting. Ended with a healthier respect for cheap shared hosting, static-first builds, and...

ADHD systems audit

A long-running effort to understand what actually helps: medication notes, planner experiments, capture systems, and how to design...

Related tools

What stayed useful

ChatGPT

The main AI workbench for everyday chat, writing, synthesis, planning, coding help, and image generation.

ChatGPT Atlas Browser

Adds browser context and agent-style capabilities to ChatGPT, which makes it more useful for live research and practical workflow execution.

Perplexity

Best for quick factual lookup and broad research, basically filling the role Wikipedia used to play for a lot of curiosity-driven searching.

NotebookLM

Excellent for grounding AI output in your own source material, especially when working through research-heavy topics or large note collections.

Anthropic

Useful as a second model perspective for reasoning-heavy work, writing, and comparing workflow quality across AI tools.

Grok

Another model in the testing mix when comparing outputs, browsing behavior, and how different tools handle practical tasks.

Adobe AI Tools

The Adobe AI features bundled into Creative Cloud are useful for image work, design iteration, and speeding up production tasks without adding another paid stack.

Sublime Text

Fast, lightweight, and still one of the cleanest editors for quick code edits, config changes, and low-friction writing or scripting work.

Lessons and offers

Carry something forward