Guide/Saving Conversations

Saving Conversations

Capture everything important from a conversation so you can pick up exactly where you left off.

Save your AI conversation to a project

When to Save

Save a conversation when:

  • You've made progress you don't want to lose
  • You're about to close the chat but plan to continue later
  • You've made important decisions you want to remember
  • The conversation contains context you'll need again

How to Save

1

Enable the Contextable connector

In your AI chat, open the connectors/tools menu and make sure Contextable is enabled for this conversation.

2

Copy and paste the save prompt

Go to Prompts, copy the "Save Insights" prompt, and paste it into your chat.

3

Approve the save

Your AI will analyze the conversation and save it. You may need to approve each tool call (this is your AI platform's security feature).

4

Confirm in your dashboard

Your new project will appear in Projects. Click it to see what was saved.

Claude saving conversation to Contextable
The AI analyzes your conversation and saves structured artifacts

What Gets Saved

When you save a conversation, the AI creates a comprehensive document that includes:

The Problem

What you're trying to solve

Goals & Success Criteria

What "done" looks like

Current Status

Progress, blockers, next steps

Key Decisions

What you decided and why

Technical Details

Architecture, constraints, terminology

Resume Prompt

How to continue later

Project detail view showing multiple artifacts with titles, types, priorities, and summaries
Your saved context, organized and ready to load

Organizing Your Saves

For larger projects, you can organize content by importance level. This helps your AI load only what's needed, saving context window space.

Core

Always loaded first

Key decisions, architecture, essential context. Mark your most important artifacts as "core" so they're always available.

Normal

Default priority

Regular notes, code, and documentation. Loaded when there's room in the context window.

Reference

Load on demand

Detailed specs, logs, verbose examples. Only loaded when specifically requested to save context space.

Pro tip: For large projects, save multiple focused artifacts instead of one massive document. For example: "Architecture Overview" (core), "Implementation Details" (normal), "API Reference" (reference). This lets your AI load only what it needs.

Tips for Better Saves

Save periodically, not just at the end

Long conversations can hit context limits. Save every 20-30 messages to make sure nothing is lost.

Use "Update Project" for updates

If you already have a saved project, use the update prompt to add new artifacts instead of creating a new project.

Be explicit about what matters

If there's something specific you want saved, mention it before pasting the save prompt: "Make sure to capture the API design we discussed."

Break up large content

Instead of one giant artifact, create multiple focused ones. This makes loading faster and lets you selectively load what you need.