CounterpartsDirectory

Guides

Guides to AI memory

What AI assistants remember, how to change what they keep, and what happens when a model retires. Each guide lists its sources.

Claude Code memory

Does Claude Code remember between sessions?

Does Claude Code remember between sessions? What CLAUDE.md and auto memory carry over, how to carry your preferences into every project, and how to give Claude a memory of you across projects.

Which memory plugin should you use with Claude Code?

Which memory plugin to use with Claude Code: 14 options, ours included, by who writes the notes, where they live, what they cost and what goes wrong.

claude-mem: CLAUDE.md files, usage costs, and where its data lives

Is claude-mem good? What it writes into your repo, what its second AI costs on your plan, where its data lives, and how to turn each one down, from its own docs.

Best memory MCP server, and how to set one up

A memory MCP server only helps when the model calls it. Which servers bring memory in on their own, where data lives, keys, clients, and setup steps.

claude-engram or Counterparts: which to use, and how to bring old notes across

claude-engram and Counterparts, compared by the people who make both: what each keeps, when staying is fine, and a manual way to bring old notes across.

How AI memory works

Why does AI memory go stale, and how do I fix it?

Why AI memory fills with stale, wrong and duplicate facts, what each memory system does when a fact changes, and how to see, correct or delete one memory yourself in Claude, Gemini, Copilot, Claude Code, Mem0, Zep and others.

What’s the difference between AI memory, RAG and the context window?

What the context window, RAG and AI memory each are, how they fit together, and when to use each, with what the research shows about long context.

Why does Claude forget between chats, and how do you stop re-explaining yourself?

A new Claude chat doesn't load your old ones. Here's what each fix carries (memory topics, instructions, a project, a pasted note) and how to give Claude a memory of you that picks up where you left off.

NotebookLM vs a chat that remembers you

NotebookLM, renamed Gemini Notebook, answers from your sources and keeps each notebook separate. When you want that, when you want an AI that remembers you across conversations, what carries over between notebooks, and how to use both.

Can an AI remind me, or message me first?

Which AI assistants can send a reminder to your phone, which only run scheduled tasks, and which hand off to Reminders or Clock: ChatGPT, Claude, Gemini, Copilot, Alexa+ and more. Plus medication reminders.

Privacy and control

What does "local" AI memory keep on your machine, and what still leaves?

What "local" means for eight AI memory systems, ours included: where the store, the search and the model run, what still leaves, and how to check it yourself.

Who can read what your AI’s memory keeps, system by system

Where 21 AI memory systems keep what they remember, who can read it (you, your workspace, the company’s staff, the model’s maker) and how to delete it, from each vendor’s own docs. Ours included.

How do I make an AI forget something, and what stays behind?

How to delete one fact, one chat or everything an AI remembers in ChatGPT, Claude, Gemini, Copilot, Perplexity, Claude Code and Counterparts (ours), what each company keeps afterwards, and how to check what it still remembers. From each vendor’s own pages; OpenAI’s consumer help pages couldn’t be read.

Model retirement

When is this Claude or ChatGPT model retired, and what carries over?

Retirement dates for Claude and OpenAI models, on the API and in the claude.ai and ChatGPT apps, what “deprecated”, “retired” and “not sooner than” mean, and what happens to your chats, projects and memory.

Claude Sonnet 4.5 retires on November 30, 2026: what stops working and what you can save

Anthropic retires Claude Sonnet 4.5 from the Claude API on November 30, 2026. What stops working, what to save first, and what carries over to Sonnet 5.5.

What’s different between Claude Sonnet 4.5 and Sonnet 5.5?

We sent the same 20 prompts to Sonnet 4.5, 4.6, 5 and 5.5, with Opus 5.5 for reference, four times each. How length, warmth, dashes, formatting and pushback differ, with real replies and the test’s limits.

More to read