I Watched Iman Gadzhi’s “Master 98% of Gemini” Video So You Don’t Waste 26 Minutes

Business/strategy angle (why Gemini, not just ChatGPT/Claude)

The creator argues Gemini is more monetizable for day-to-day work because it is embedded in Google’s workflow tools, such as:

• Gmail
• Docs
• Sheets
• YouTube
• Chrome
• Search

This reduces adoption friction—you’re not asking people to “start from zero” in a new AI app every time.

Core thesis: Pick the AI tool that integrates where work already happens, so you capture value continuously instead of switching tools and re-establishing context.

---

Key “playbook” for using Gemini (execution framework)

Tip 1: Personal intelligence setup

Build an “intelligence layer” using:

• Memory (import prior context)
• App connections (link calendar + Google apps for real context)
• Instructions (define behavior constraints)

Migration approach (from ChatGPT/Claude)

• Gemini gives a prompt you can paste into existing AI workflows to get a summary of what Gemini should “remember.”
• Optionally, export chat history and upload it to Gemini.

Guardrails

• Instruct Gemini not to be a yes-man; it should push back when ideas are weak.

Time investment

• Spend ~15 minutes on setup so the system pays off daily.

---

Tip 2: Model selection = token/cost control

Gemini has three model tiers:

• Flashlight: low-effort tasks
  • e.g., summaries, quick organizing, light back-and-forth
  • described as barely touching your tokens
• Flash: daily-driver
  • reasoning without heavy token burn
• Pro: high-stakes/urgent work
  • when something is due “yesterday”
  • described as burning tokens fast → use sparingly

Token economy principle: context is treated like “currency,” so it must be managed carefully.

Cost anecdote (organizational KPI framing)

• The presenter claims the company spends hundreds of thousands per month on tokens, positioning token management as an operational necessity.

---

Tip 3: Reuse via Gems and Skills (reduce repeat prompting)

Problem

Users repeatedly type the same instructions to AI.

Solution

• Gem: a customized chat + knowledge base
  • Example: “how you write emails” → build once, reuse.
• Skill: a repeatable process that can be called automatically
  • Example: describe what you want; Gemini generates it, or upload a prior “skill file.”

The summary notes a library of thousands of ready-made Gems/Skills.

---

Intermediate workflows: Canvas + “live editing”

Canvas keeps:

• chat on the left
• work/output on the right

…and updates in place, reducing the need for copy/paste of large outputs.

Examples:

• Turn a book chapter into a 10-question interactive quiz
• Build/edit assets like workout tracker apps
• Edit scripts through structured collaboration

---

Omni-screen capability (no copying)

In Chrome, Gemini can see the screen, so you can ask questions about a page without copying text.

Example workflow: open multiple tabs and use the “@” approach in tasking to focus attention.

---

Advanced automation: Spark / Deep Research

Spark

AI runs tasks proactively “while you sleep.”

• Example: summarize your inbox overnight into a Google Doc

Deep research

An automated research pipeline where Gemini:

1. builds its own prompt
2. interviews you with questions
3. produces a finished report

Then it outputs into NotebookLM audio overview for podcast-style consumption.

---

Metrics / KPIs / targets mentioned (high-level)

• Setup target: spend ~15 minutes configuring Gemini.
• Cost signal (token spend): “hundreds of thousands a month” (token spend framed as a key operational constraint).
• Model usage guidance (behavioral targets):
  • Flashlight for minimal reasoning tasks
  • Flash for most daily work
  • Pro sparingly for urgent/high-consequence output

(No explicit revenue, CAC, LTV, churn, or time-to-value metrics were stated beyond the token-cost anecdote and efficiency claims.)

---

Concrete actionable recommendations (as stated)

• Import memory + chat history from ChatGPT/Claude to avoid “starting from zero.”
• Link calendar + Google apps so Gemini has real work context.
• Add instruction to challenge weak ideas (avoid yes-man behavior).
• Separate tasks by model tier to control token burn.
• Build reusable Gems/Skills so you stop rewriting prompts.
• Use Canvas for structured iterative production (e.g., quiz generation from a chapter).
• Use Spark and Deep research to automate background work (e.g., inbox → Doc; research → NotebookLM audio).

---

Caveats / positioning verdict

The creator’s critique: Gemini is “good at everything and great at nothing.”

Reason given: Google deployed Gemini widely across products, which can reduce depth or exceptional performance in any single workflow due to rapid changes and broad surface area.

---

Presenter / sources

• Iman Gadzhi (original creator whose video and claims are referenced)
• The summarizer/presenter (channel not named in the subtitles; the narrator who summarizes the video)