Almost everyone runs Claude exactly as it shipped. The defaults are fine. But four settings, none of which are on by default in the way that matters, quietly change how good the tool is at being your tool.
None of this takes longer than a coffee. I have ordered them by how much they change your day. A quick, honest caveat before we start: Anthropic ships changes to these menus often, and one of them was renamed recently, so I have used the current labels as of July 2026 rather than the ones you may have seen in older screenshots. If a name looks slightly different in your account, the description will still match.
One: Turn On Memory
Go to Settings, then Capabilities. This is where memory lives, and it is worth knowing that it is actually two separate features, not one switch. People conflate them and then get confused about what is on.
- ✓Search and reference chats. Lets Claude look back through your earlier conversations and pull from them in a new one. This one is on paid plans (Pro, Max, Team, Enterprise), and on those plans it is generally on already.
- ✓Generate memory from chat history. This is the persistent memory that builds a durable profile of how you work over time. It is opt-in. You have to switch it on yourself, and most people never do.
The second one is the point. With it on, Claude remembers your context across sessions, so you stop opening every chat by re-describing your business, your stack, and your preferences. In the same panel you can view and edit your memory, pause it without deleting anything, or reset it entirely. If you are privacy-conscious, that edit view is the thing to bookmark: memory is not a black box, you can read exactly what it has retained.
Do not skip the edit view. Turning memory on is step one. Spending two minutes in "view and edit your memory" to correct or delete anything wrong is what makes it trustworthy. A memory you can audit is a memory you can rely on.
Two: Write "Instructions for Claude"
This is the one that is criminal to skip, and also the one whose name trips people up. It used to be talked about as "Preferences." In the current app it is called Instructions for Claude, reached from Settings. It is an account-wide box that shapes every conversation you have.
Left blank, Claude writes in a competent, slightly corporate default voice. Fill this box in once and it stops sounding like a press release and starts sounding like you. Paste something like this and adapt it:
Write the way I talk: direct, plain, no filler. Short sentences. No "delve", no "in today's fast-paced world", no bullet lists unless I ask. Lead with the answer, then the reasoning. If I'm wrong about something, say so plainly. Ask me a question back only when you genuinely can't proceed without it.
Two things worth knowing here. First, the more specific you are, the better it works. "Be concise" is weak; "no sentence longer than 25 words, no adjectives you would not use out loud" is strong. Second, if you have used the old Styles feature (Concise, Explanatory, Formal), note that Anthropic is retiring those and moving that behaviour into Skills. If a style you relied on disappears, the fix is to describe the same behaviour right here in your instructions.
A blank instructions box is you letting the model guess who you are. Two paragraphs and it stops guessing.
Three: Stop Typing. Start Talking.
You speak roughly three times faster than you type, and most of the friction of using an AI well is the cost of writing a good, long prompt. Talking removes it. Here is the part the "go download an app" advice usually misses: Claude already has dictation built in.
Tap the microphone
The Claude iOS and Android apps have a mic icon in the chat bar. Tap it, talk, tap send. It is available on every plan, including Free. Audio is transcribed and then discarded.
Press Caps Lock
The Claude desktop app for Mac has Quick Entry dictation: press Caps Lock to dictate with live transcription. It is off by default, so enable it first in Settings, General. Voice dictation needs macOS 14 or later.
Built-in dictation covers the everyday case completely, and it is free. A dedicated third-party dictation app is still worth it if you want system-wide voice input across every app you use, not just Claude, or if you want custom vocabulary and punctuation control. But try the built-in option first. Most people never realise it is there and pay for something they already had.
Do not confuse dictation with voice mode. Dictation is speech to text: you talk, it types, you read the reply. Voice mode is a two-way spoken conversation where Claude talks back. Both exist. For drafting prompts fast, you want dictation.
Four: Keep Anthropic Academy Open Beside You
This is not a setting inside the app, it is a habit. Anthropic Academy is Anthropic's own course library, and it is free. You will find it at anthropic.com/learn.
It runs from non-technical foundations through to building on the API, with short self-paced courses like Claude 101 and Claude Code 101, and it hands out completion certificates. The way to actually use it is not to sit and binge a course. Split your screen: Academy on one side, a live Claude chat on the other, and try each thing the moment you learn it. You retain almost nothing watched passively and almost everything you immediately do.
The Bonus: Switching From ChatGPT? Bring Your Memory.
If you are moving over from ChatGPT, Gemini, or Grok, you do not have to start Claude from a blank slate. There is a native memory import tool at claude.com/import-memory, also reachable from Settings, Capabilities, Memory.
It works in a slightly clever way: Claude gives you a prompt to paste into your old assistant, which dumps out everything it knows about you, and you paste that dump back into Claude's import box. Claude turns it into structured memory over the next day or so. Anthropic labels the feature experimental, so check the result in the memory edit view and re-add anything important that did not carry over. But it means a switch costs you minutes, not months of rebuilt context.
Five Minutes, Then Never Again
- 01Memory on in Settings, Capabilities. Turn on "Generate memory from chat history," then read what it saves.
- 02Instructions for Claude filled in with how you actually want it to write.
- 03Dictation tried first as the built-in feature, on mobile or Mac, before paying for anything.
- 04Anthropic Academy open in the next tab, learned by doing, not watching.
None of these are hidden, exactly. They are just off the path most people walk. Spend the five minutes once and every session after it starts from a better place.
Want this kind of agentic setup for your own brand or product?
I help founders and brands build generative AI systems that actually ship: content engines, agentic workflows, and AI-built sites that are production-grade down to the schema layer. If you want a second pair of eyes on your setup, book a session.
Book a Growth Chat
Settings paths and feature names in this post reflect the Claude apps as of July 2026 and were checked against Anthropic's own help documentation, including the pages on chat search and memory, personalization and instructions, dictation, and memory import. Anthropic updates these menus frequently and labels memory import as experimental, so treat the exact wording as current-at-writing and confirm against your own account. Nothing in this post is sponsored.