Organizing Conversations, Titles, and History

As you use StretchGPT every day, your conversation list becomes a working history. This guide covers how StretchGPT titles and remembers conversations, and how to keep them organized so you can always find the thread you need.

How conversations are stored

Each conversation is a saved thread of messages tied to your account and organization. StretchGPT records the family, the title, a running message count, and when the conversation last had activity. Your list is sorted with the most recently active conversation first, so whatever you're working on stays at the top.

Every conversation is strictly isolated: one account can never see another account's conversations, and conversations don't leak across organizations.

Titles: automatic and manual

When you start a conversation, StretchGPT gives it a title automatically:

• If your first message is short, the whole message becomes the title.
• If it's longer, the title is the first part of the message, trimmed to about 60 characters with an ellipsis.
• If there's no first message yet, the title is simply New chat.

You can rename any conversation to something clearer:

1. Open the conversation or its menu in the list.
2. Choose Rename.
3. Type the new title and confirm.

The memory window

StretchGPT gives the model the recent history of a conversation — up to the last 20 completed user/assistant turns — as context for each new answer. A few practical consequences:

• Follow-ups work naturally within a thread because recent turns are always in context.
• Very old turns age out. In a long-running conversation, details from dozens of turns ago may no longer be in the active window. If an old fact matters, restate it.
• Only completed turns count. Blocked or errored messages are excluded from history, so they never confuse later answers.

Archiving and cleanup

When you're finished with a conversation, archive it to remove it from your active list:

1. Open the conversation's menu.
2. Choose Archive (or Delete).
3. The thread leaves your active list and no longer appears in your history view.

Archiving is the tidy way to keep your list focused on current work without permanently losing a thread's record.

A realistic example

You run three ongoing threads: a code thread named "Refactor export job," a research thread you renamed to "Q3 competitor scan," and a Core thread still auto-titled "Draft welcome email for new hires…". Because the list sorts by recent activity, whichever you touched last sits on top. When the export refactor ships, you archive that thread so it stops cluttering the list — but the other two stay exactly where they were.

Tips

• Rename as soon as a thread matters. A 5-second rename saves a frustrating search later.
• One conversation per deliverable. Focused threads keep the 20-turn memory window on-topic and produce better answers.
• Archive aggressively. A short active list is a fast active list.
• Restate key facts in long threads. If you've gone well past 20 turns, remind StretchGPT of the crucial detail rather than assuming it's still remembered.

FAQ

Does StretchGPT ever read my other conversations for context? No. Context comes only from the current conversation's recent turns.

Can other people in my organization see my chats? No. Conversations are private to the individual account that created them.

What exactly is the memory limit? StretchGPT feeds up to the last 20 completed user/assistant turns from the current conversation into each answer. Blocked and errored messages don't count.