Eight Newer Claude Features, Tried on Real Work · Lesson 1 of 8
Teach Claude your business once
15 minutes. Works on: all plans. On Team and Enterprise plans, your owner decides whether memory is available, and it stays off for you until you turn it on.
---
Here's a habit most people have without noticing. You open a new chat and type the same three sentences you typed yesterday. "We're a heating and cooling company. We serve these four towns. Keep it friendly and short." Then you ask your question.
Claude can remember that for you now. This lesson shows you how to teach it the basics once, check what it remembered, and fix or delete anything that's wrong.
Memory isn't brand new, but it got much easier to see and control in 2026:
- March 2, 2026: memory came to everyone, including the free plan.
- July 10, 2026: memory became a list of separate entries that Claude reads and updates as you talk, instead of one daily summary.
- August 25, 2026: everything Claude remembers is now listed under Topics in Settings > Memory, where you can read, edit or delete any of it.
On the free, Pro and Max plans, memory is on by default. On Team and Enterprise plans, the owner controls whether it's available, and it's off for each person until they turn it on.
Memory is for the things that are true about you and your business every day:
- What you do, and where.
- Your hours and how customers reach you.
- How you like things written.
- The words your business uses ("service call," not "ticket").
It is not the place for things that change every week, like a price list or this month's schedule. Those belong in a file you attach or a Project. (There's a tip on that: [Give Claude your price sheet once, in a Project](/tips/project-for-your-price-sheet).)
Anthropic already keeps some things out. By default Claude doesn't store sensitive personal topics such as health, politics or religion. And some things it says it never saves, even if you ask: government ID numbers, criminal history, financial account numbers and immigration status.
Add your own rule on top: no customer details, no passwords or door codes, nothing about a coworker you wouldn't say to their face. Memory is about your business, not about the people in it.
You don't fill in a form. You just tell Claude, in a normal chat.
At Comfort Zone, Tina typed this:
> Please remember these things about my work. I'm the office manager at Comfort Zone Heating & Cooling, a 20-person heating and air conditioning company. We serve four towns: Millbrook, Easton Falls, Carver and Pine Hollow. Office hours are 7:30 to 5, Monday to Friday, with an emergency line after hours. We call our jobs "service calls" and our maintenance plan "the Comfort Club." When you write to customers, keep it short and friendly, no exclamation points, and sign it "The Comfort Zone team."
Claude saves what it thinks is worth keeping. Anthropic says changes apply from your next conversation.
You can also tell it to change or forget something the same way: "Forget that we serve Carver. We stopped going there in September."
- Open Settings > Memory.
- Look at the list under Topics.
- Select any topic to read it.
- Use the edit icon to fix it, or Delete to remove it.
Read it like you'd read a note a new hire wrote about your business. Is it right? Is anything in there that shouldn't be?
One thing to know: deleting a chat doesn't delete the memories that came from it. If you pasted something into a chat by mistake, delete the chat *and* check Topics.
One chat without memory. Starting a chat that has nothing to do with work? Before your first message, click + and turn off Memory for that chat. You can't change it after the first message.
Incognito. The ghost icon in the top right starts a chat that isn't saved to your history or to memory. On Team plans it's still kept for a while and included in company data exports, so it's tidy, not secret. ([Claude for Your Team](/courses/claude-for-your-team) covers this in Lesson 2.)
Projects keep their own memory. Each Project has its own memory, so what happens in your "Customer replies" project doesn't leak into your "Hiring" project. A Project also uses your general memory unless you turn off Use account memory on the Project's page.
Bonus, if you're switching from another AI: under Settings > Memory there's an import option (Start import). It gives you a prompt to run in the other tool and paste back. Anthropic calls memory import "experimental," so read what it saved afterward and delete anything personal.
In Settings > Memory you can turn off Generate memory from chats. You'll then choose Pause memory (keeps what's there but stops using it) or Reset memory (deletes everything, permanently). Pause if you're unsure. Reset only if you really mean it.
- Open the Memory Starter List template and fill in five facts about your business. Leave the "never" column filled in too.
- Paste your five facts into a new chat with "Please remember these things about my work."
- Open Settings > Memory > Topics. Fix one thing that's slightly off. Delete one thing you don't need it to know.
- Start a new chat and ask: "A customer in Pine Hollow asks if we can come Saturday. Draft a two-line reply." (Use your own town and question.)
Check it worked
- The reply used your service area, hours and sign-off without you typing them.
- Topics shows only things about your work. No customer names, no codes, nothing personal about a coworker.
- You know where to pause memory and how to start one chat without it.
---
Next: ask for a chart instead of a paragraph, and fix a draft one piece at a time.
→ Lesson 2: Ask for a picture, then fix it in place