Module 01 · Foundational Mindset
Comparing Claude, ChatGPT, and Gemini Without the Hype
Open lesson + course map
On this lesson
Course outline
Module 1 · Foundational Mindset
Module 2 · Mastering Context Threads
Module 3 · Building Custom GPTs and Gems
Module 4 · Multi-Model Workflows
Module 5 · From User to Operator
You do not need a permanent winner between Claude, ChatGPT, and Gemini. You need a repeatable way to compare whichever versions and plans you can access for the task in front of you. The useful mindset shift is from “chatting with a bot” to briefing a tool, checking its output, and keeping a human responsible for the result.
// concept
Start With the Job, Not the Brand
These products change frequently. Features can differ by plan, region, device, account type, and workspace policy. Treat the table below as a set of experiments to run, not as a permanent ranking:
| Tool | A useful test task | What to verify first |
|---|---|---|
| Claude | A document-heavy draft or a coding task | Whether Projects, uploads, and the needed usage limit are available on your plan |
| ChatGPT | General drafting or a configurable GPT workflow | Whether GPT creation, the model, and the required tools are included in your plan |
| Gemini | A task using Google services or mixed media | Whether connected apps, the model, and the needed limit are available on your account |
The practical rule is simple: run the same representative task in the tools you can legitimately access, score the outputs against the same checklist, and use the tool that reduces your verified editing work. Never delegate final responsibility for client, legal, financial, medical, or safety-sensitive work to a model.
// concept
From “User” to “Operator”
A casual user asks a question and accepts the first response. A careful operator gives the model a role, relevant facts, a clear task, and an output format, then reviews the result.
The Four-Part Brief
- Role — the useful perspective for this task, without inventing credentials.
- Context — the facts, audience, constraints, and source material the model needs.
- Task — one clear action and a definition of done.
- Format — the structure, length, language, and fields you want back.
Role: Act as a copy editor for a Karachi-based ecommerce brand.
Context: We are launching a wallet priced at PKR 3,500. The verified product
facts are: full-grain leather and hand-stitched in Korangi.
Task: Draft three Instagram caption options. Do not add product claims.
Format: Numbered list; each caption under 40 words; one emoji maximum.Compare that with “write me an Instagram caption for a wallet.” The structured version gives you more specific criteria to inspect, but it still needs fact-checking and editing before publication.
// concept
Score Outputs Instead of Picking a Favourite
Use the same five-point review for each tool:
| Check | Pass condition |
|---|---|
| Facts | Uses only facts you supplied or sources you can verify |
| Instructions | Completes the named task without adding a different one |
| Format | Meets the requested structure and length |
| Safety | Does not expose private data or invent risky advice |
| Editing load | Requires the least correction while preserving your voice |
Record the plan and date alongside your result. A test from a free account this month does not prove how a paid workspace or a later model version will behave.
// concept
Context Windows Without the Mythology
Every model has a context window: a limit on how much input it can use together. Product interfaces may also summarize, retrieve, or omit earlier material. If an important constraint stops appearing in the output, re-supply it and test again instead of guessing why it was missed. Module 2 covers a deliberate handoff method.
Do not paste confidential client material merely to make the context richer. Remove personal data, secrets, private contracts, and credentials unless you have authorization and the account’s data controls meet the work’s requirements.
// pakistan_angle
Pakistan Angle
A USD-priced subscription can become a meaningful PKR expense after exchange-rate and issuer charges. Start with tools and limits available to your own account, record where a limit actually blocks a real task, and check the vendor’s current plan page before spending. Payment eligibility varies by provider, country, and bank; use only a payment method that the vendor and issuer explicitly support. This course does not recommend card or regional-payment workarounds.
Low-bandwidth practice is still useful: test with a short, representative excerpt before uploading a large file, save important briefs locally, and keep a manual route for time-sensitive work.
// hands_on
Do This Now
Open any two or three assistants that are legitimately available on your accounts. Give each the same four-part brief:
Role: Act as a career-coaching assistant for an early-career professional.
Context: I have one year of digital-marketing experience in Pakistan and want
a role that uses AI tools. Do not invent experience or job openings.
Task: Suggest three job-title search terms and explain why each may fit.
Format: Exactly three numbered items; one sentence per item.Score each response with the five checks above. Save the date, plan, and result. The goal is not to crown a universal winner; it is to learn which available tool fits this task with the least verified editing.
// sources
Sources and current-plan checks
3 official sources — check every claim yourself
Self-check
Before you mark Lesson 1.1 complete
- Can I explain “Comparing Claude, ChatGPT, and Gemini Without the Hype” without reading the lesson back word for word?
- Did I complete the lesson’s practice step on a real or clearly labelled sample task?
- Did I check the result for invented facts, private data, unsafe actions, and mismatch with the brief?