Free to read · no account needed

14 min · foundations

Path: Your first 7 days with AI chat (~14 min/day)

Goal: Turn the catalog into a one-week journey: safety → clear prompts → real work → verify → pick defaults.

Best for: Absolute beginners who want a sequence, not a random pile of articles.

Why beginners skip this

Cramming the seven-day plan into day one feels bold—day three will be empty. One real artifact per day.

Why this order (not random clicking)

Day 1 locks the account before you paste work; days 2–3 practice reusable templates on real tasks; day 4 learns to distrust fluent wrong answers.

Skip Agents, MCP, and plugins on purpose this week. Those pages assume you already have a default chat habit.

Outcome after day 7

  • One default chat app with MFA on and a personal never-paste list.
  • A Goal · Format · Boundaries template you can fill from memory.
  • Three real tasks finished (email, notes, or study)—not toy demos.
  • A habit: verify numbers/quotes before you share.

A week in someone’s shoes

Scene: Jordan (ops coordinator in HK) keeps bouncing between three AI apps and still sends stiff emails. This week Jordan picks ChatGPT as default, writes a never-paste list on Day 1, rewrites one real vendor reply on Day 2, verifies a “confirmed Friday” claim on Day 4, and on Day 7 keeps ChatGPT for daily mail while bookmarking Claude for long polish.

Example

Filled prompt

Jordan Day 2 (18 minutes):
Goal: Reply so the vendor knows we need the corrected invoice before Thursday payment.
Format: ≤100 words; one ask.
Boundaries: Do not invent fee amounts.
Draft: “hey can you fix the invoice thx”
→ one warmer rewrite → Jordan checks the Thursday date in the finance sheet → send.

What good output looks like

  • Real owed reply, not a fictional Acme demo
  • One edit pass, not ten tool switches
  • Human owns the final date

Day-by-day plan (~20 minutes)

DayDo thisRead / use
1Create/secure account; write never-paste liststart.html · safety-secrets
2Send one GFB prompt on a real email/notes taskprompt-goal-format-boundaries · packs A/D
3Learn what chat can/cannot do; try one study or writing packfoundations-what-an-llm… · packs
4Practice “sure but wrong”: verify one answer with the 1-min methodfoundations-when-ai-sounds-sure-but-wrong
5Second real task (meeting notes or weekly update)claude-meeting-notes · writing-stakeholder-update
6Same prompt on your tool + one other; keep the better editcompare.html
7Keep or switch default; pin template; pick a pathlearn.html paths
Example

Filled prompt

Day 4 verify drill (12 minutes):
1) Ask your default tool: “Summarize whether our Friday demo is publicly confirmed.”
2) Run the 1-minute ritual: underline numbers/dates/claims the model added.
3) Open one primary source (calendar invite or project doc)—or delete the claim.
4) Paste only what you can defend.

What good output looks like

  • Model-added ‘confirmed’ is challenged
  • Primary source wins over fluent chat
  • Nothing ships unverified

Rules that keep the week short

  • If you miss a day, continue; do not restart from Day 1.
  • One default tool all week; Day 6 is the only comparison day.
  • Optional stretch after Day 7: path-writing, path-career, path-study, or path-office-weekly-routine.
Example

Filled prompt

Live artifact this week for: Turn the catalog into a one-week journey: safety → clear prompts → real work → verify → ch
Name real files/people. No invented facts; label Unknowns; stop before send/delete.
Return working prompt + one pass/fail check I can mark in a minute.

What good output looks like

  • Calendar slot titled with slug for “Turn the catalog into a one-week journey: safety →”
  • Live artifact opened — not only the bookmark
  • Practice timed; refuse-line written
  • Pass/fail: artifact dated Asia/Shanghai
Example

Filled prompt

Day 6: one real GFB email you owe. Save dated winner.
No third tool install. Stack card still one default.
Context: real workstream this week (Asia/Shanghai). Name actual files/people.
Constraints: no invented metrics/owners/prices; label Unknowns; stop before send/delete/pay.
Done check: I can mark pass/fail in one minute on the output.

What good output looks like

  • Live artifact opened — not only the bookmark
  • Practice timed; refuse-line written
  • Pass/fail: artifact dated Asia/Shanghai
  • Calendar slot titled with slug for “Turn the catalog into a one-week journey: safety →”

Red flag

Cramming seven days into day one empties day three—one real artifact per day.

Save this card

Card: today’s day number + one practice checkbox. Open the live day card, not seven titles.

Before you start

Create a 25-minute calendar block titled path-first-7-days. Open the live artifact for “Turn the catalog into a one-week journey: safety → clear pro”—bookmark alone does not count Keep the live artifact visible the whole time. If you only have a blank box, stop and fetch the real job first

How to use this week

This week ship one real artifact each day along the seven-day path; no checkmark without an artifact that day.

Pitfalls

PitfallFix
Jumping to agents on day 2Finish GFB + verify first
Paying before a felt limitFree until blocked
Skipping safety listDay 1 deliverable

Practice now

Accountability ping: tell one person day-N done (Asia/Shanghai).

Habit streak: mark day N complete only with artifact path (Asia/Shanghai).

Day checkbox: tick today’s live task before sleep (Asia/Shanghai).

7-day drill: day-N checkbox—one live task done + one verify habit. Fake tasks fail the week (Asia/Shanghai).

Thin-spot drill: rewrite the Save card with real nouns from your work (repo, doc, inbox). Generic placeholders fail the drill.

Tonight ship one real artifact (prompt, note, or checklist); no next day without it.

Close the loop

Write one line: next time I cram seven days into one, I refuse empty slots.

One sentence takeaway

One real artifact per day; no checkmark without it that day.

Next

Start · Packs · Compare · Mistakes · Accounts and safety

← Back to home · Learn · Tools · Compare · Start