🎤 AI Show & Tell #1 · Wix Marketing · 40 MIN

How to Never Hit Your Claude Session
Limit Again

Context Window 4%
↺ reset early
Fill → reset early → never hit the limit. ▲ reset here
Part 1 — The Problem

What is a token?

A token ≈ ¾ of a word. Every new message makes Claude re-read the entire conversation — so the cost of each turn compounds fast.

500
Message 1
5,000
Message 10
×10
15,500
Message 30
×3

98.5% of those tokens are just Claude re-reading history — every single turn.

Part 1 — The Problem

AI Dementia

As the context window fills, Claude gets worse at finding the right details — even though they're all still in the window. That's context rot.

The double penalty: worse recall → you re-ask and re-explain → even more tokens burned.

Retrieval accuracy vs. tokens in context
95%
64k
92%
256k
85%
512k
78%
1M
Part 1 — The Problem

The Context Zone

Sharp Zone
0–25%
Drift Zone
25–50%
Dementia Zone
50–100%

Most people work in the red. ↑

Part 2 — The Fix

Reset early. Store the rest.

Reset Protocol

01

Ask Claude for a summary of everything decided so far

02

Open a new chat

03

Paste the summary back — continue from there

Memory Storage Options

Native
Claude Projects
Built-in memory for a body of work. Ask Claude to save decisions — also happens automatically.
External
Obsidian
Your own external knowledge base.
Plugin
Claude-mem
Persistent memory, automatic.
Part 2 — The Fix

You might be asking yourself…

How do I check context usage in Claude.ai or Cowork?

You can't. No %, no meter, no indicator.

Read the signals instead

Claude contradicts its earlier answers
Replies feel generic or misaligned with context
It forgets something you said 30 messages ago
The session feels slower or less precise

Don't try to monitor it. Build the habit of resetting early by default — any time the task shifts significantly, open a new chat.

Part 2 — The Fix

One task, one chat.

Don't mix jobs in a single session — each one drags its history into the next. Research → Write → Review: one job per chat = small, sharp context.

Pick the right model

Haiku

Fast & cheap · simple tasks

Sonnet

Balanced · your default

✓ DEFAULT

Opus

Heavy reasoning · high-stakes only

Part 3 — Going Deeper · Claude Code

Claude Code: the toolkit

Slash commands for power users:

/clear

Wipe the session

/compact

Summarize & shrink the session

/re

Drop a failed attempt

/btw

Side question, off-thread

/context

Check how full you are

CLAUDE.md <200 lines, max. Every line costs tokens — every session, forever.

Part 4 — The Advisor Strategy

A junior with a senior on call

The mistake: using Opus for everything — like hiring a senior consultant to answer your emails. Expensive and unnecessary.

Worker

Sonnet / Haiku

Runs every step — drafting, researching, formatting, iterating. Constant, and cheap.

Advisor

Opus

Never runs by default. Called only when stuck — hard reasoning or a judgment call. Weighs in, then the Worker continues.

🎁 Your Gift

Two things to take home

01

18 Claude Tricks You Wish You Knew Sooner

A reference card — save it, pin it, keep it close.

Poster / Reference Card
02

The Memory Extraction Prompt

Paste into ChatGPT or Gemini — it extracts everything it knows about you — import that context into Claude.

Copy-Paste Prompt
AI Show & Tell #1

Questions?

Reset early. Store the rest. Never hit the limit.