# AI 101 · Lesson 4 — Give it your context

**The CV Method · AI 101 · Lesson 4 · Working with AI · Aug 2026**
*by Corina · @the_cv_method · hello.thecvmethod.com · current as of Aug 2026 — the AI space moves fast, so double-check tool names*

The gap between "generic AI" and "this sounds like **me** / knows my business" isn't a better model — it's context. Here's how to feed it your world once, so every answer lands.

---

## The steps (at a glance)

1. **About doc** — who you are, what you do, voice + rules
2. **Gather** — examples, brand voice, products, FAQs — one folder
3. **Point** — tell it to read your context first
4. **Voice** — paste 2–3 things you've written
5. **Keep clean** — update as you go; keep secrets out

## Rule zero — Context is the difference.

Same model, two answers: generic when it knows nothing about you, sharp when it knows your business, your voice, and your rules. This lesson is how you give it that — once.

## Step 1 · Write your 'about' doc

Put the basics in one file (your **CLAUDE.md** from Lesson 1): who you are, what you do, who you help, your tone, and your hard rules. This is the note the AI reads before it answers — so it never starts from zero again.

*The move: have the AI interview you to build it — 5 questions beats a blank page.*

## Step 2 · Gather the real reference files

Collect the real docs it needs — past examples, your brand voice, a product list, FAQs — in **one folder**. These are the source of truth. The AI uses them instead of making things up, which is the whole point.

*The move: start with the 3 files you re-explain most often. Add more only when a task needs it.*

## Step 3 · Point it at them — every time

Tell it, up front, to **read your context before answering** and to follow it over anything generic. This is the difference between a doc that sits there and a doc that actually shapes the output.

*The move: if you set up Lesson 1, this becomes automatic in Claude Code — it reads your CLAUDE.md on every run. In the Claude app or ChatGPT, paste it in or put it in a Project's instructions.*

## Step 4 · Give it your voice by example

Paste **2–3 things you've actually written** — a real email, a caption, a note. Examples teach your tone far better than adjectives. This is what makes it stop sounding corporate and start sounding like you.

*The move: ask it to point out anything in a draft that would 'break' your voice.*

## Step 5 · Keep it current + private

Update the context as things change — stale context quietly makes it wrong. And keep **secrets and client info out**: no passwords, keys, or private details in anything the AI (or a public post) can see.

*The move: have it scan your context for anything sensitive before you use it anywhere.*

## Keep it honest & safe

- Context is the difference
- Keep **secrets & PII out**
- **"unknown"** over guessing
- Update as you go
- **Date it** — AI moves fast

## Steal these prompts

**About doc**

```
Interview me to build an 'about my business' doc: what I do, who I help, my tone, and my hard rules. Ask me 5 questions, then draft it.
```

**Gather**

```
Here are my reference docs: [paste / attach]. Use these as the source of truth — don't invent details that aren't in them.
```

**Point**

```
Before you answer, read my about-doc and reference files. Follow them over any generic advice.
```

**Voice**

```
Here are 3 things I wrote: [paste]. Match this voice — not corporate, not hype. Point out anything that would break it.
```

**Keep clean**

```
Scan my context for anything that looks like a password, API key, or private client info, and flag it so I can remove it before this goes anywhere.
```
