Guide: Migrating from ChatGPT to Change Agent – Change Agent AI

Why Make the Switch?

Before we dive into the how, let's talk about the why. Change Agent isn't just another AI chatbot—it's built specifically by and for organizers, advocates, and mission-driven, people-powered organizations.

Key Advantages Over Mainstream AI:

Movement Tool Ecosystem: Native integrations with Action Network, ActBlue, LegiScan, Gamma, and more
The Privacy You Need: Your data is yours. We never sell, share, or train on your data. Delete it, and it's destroyed within hours.
The Greener Choice: 250 users of Change Agent use roughly the same energy as running a modern refrigerator
A Model for the Movement: Our model centers impact and inspires action without perpetuating harmful bias like racism and misogyny
Staunchly Pro-Worker: We give changemakers AI superpowers to maximize impact, not profit margin—and that takes people, not robots

You can evencan move over your chat history, customized models/GPTs, and personalization from other AI tools. No need to startg over.

And beyond that, Change Agent meets or exceeds performance and features offered by the mainstream providers. No tradeoffs, just the advantages of using a platform built by and for the progressive movement.

Pre-Migration: Extract Your ChatGPT Knowledge

Step 1: Document Your General ChatGPT Personalization and Custom GPTs

ChatGPT customizes your experience in two ways: Personalization and Custom GPTs, though it’s possible you haven’t been using your own Custom GPTs.

General Personalization:

ChatGPT tracks information about everything you say and do inside ChatGPT. If you want to extract the useful information from this for Change Agent to have the same information (without using it to target you with ads), send this prompt to ChatGPT and save the response to a google doc to save. Later, we’ll upload this information into Change Agent to make it just as personalized as ChatGPT for you.

Click to Expand: Copy This Text

ROLE
You are a senior AI cognitive migration architect and knowledge
systems engineer.

Your task is to analyze my complete ChatGPT conversation history
and personalization data and reconstruct it into a structured,
high-signal personalization knowledge base that preserves identity,
context, memory, preferences, ongoing projects, reasoning patterns,
and nuanced decision history.

Analyze your entire context, memories, and personalizations for
me and produce a structured knowledge base only formatted as
bullet point entries of key "memories" about me. Each memory
will be a fully self-contained bit of information. There are
the types of things you should include:

1. Identity Profile (for user as well as the organization they
work for)
* Core personality traits relevant their AI usage
* Communication preferences
* Values & Mission

2. Long-Term Vision & Goals
* Explicit goals
* Implicit goals inferred from behavior
* Recurring ambitions 
* Strategic direction shifts over time, both personally and organizationally

3. Active & Recurring Projects
For each project:
* Name (if inferable)
* Purpose
* Current state
* Open loops
* Key decisions made
* Constraints
* Dependencies

4. Knowledge Domains & Expertise Map
* Areas of high competence
* Areas of learning
* Tools frequently discussed
* Technical stack (if relevant)
* Conceptual frameworks used
* Areas of common work (e.g. messaging, policy, tech, hr, etc)
* Repeated questions or themes

5. Decision-Making Patterns
* Risk tolerance
* Optimization style (speed vs precision)
* Strategic vs tactical bias

7. Constraints & Operating Conditions
* Time constraints
* Resource limitations
* Skill gaps
* Environmental or structural limits

8. Preferred Output Styles
* Formatting preferences
* Tone preferences
* Level of detail desired
* Structural expectations

9. Open Loops & Unresolved Threads
* Ideas mentioned but not developed
* Projects started but paused
* Strategic questions not fully resolved

RULES
* Do NOT summarize chronologically
* Do NOT restate entire conversations
* Do NOT include filler
* Preserve nuance where it affects long-term behavior or strategy
* Distill repetition into patterns
* Infer implicit traits when strongly supported by behavior
* Do NOT return statements significantly judgmental of or potentially
offensive to the user.
* Every memory bullet point must be self-contained, concise information.
For context, know that every bullet point will in the future
be included in an LLM's context without any addition information
(e.g. the numbered sections I used above). 

OUTPUT FORMAT
You must ONLY return a basic bullet point list (no sub-bullets)
of memories to store.

Here are some examples for reference:
- User's name is Pramila Jayapal. She is the U.S. representative
from Washington's 7th congressional district since 2017
- User was formerly the chair of the Progressive Caucus, but
stepped down recently
- User prefers assistant to adopt a supportive coworker/mentor
tone: direct and precise with feedback, proactively recommending
alternative solutions to problems even when not explicitly asked.
- User is interested in recent events related to AI for Progressive
politics
- User is working on a new legislative advocacy campaign in Congress
relating to Data Center construction and regulation
- User preference: When asked to draft content (emails, LinkedIn
posts, etc.), always reference memory/personalization and messaging-related
knowledge/documents thoroughly to include personalized tone,
messaging, and information

Custom GPTs:

If you’ve used Custom GPTs you want to move over, for each Custom GPT you use:

Open the Custom GPT in ChatGPT
Copy the instructions (system prompt) to your Google Doc
Save all attached files
Screenshot settings and capabilities

Step 2: Export Your ChatGPT Data

This is necessary if you want to import your chat history from ChatGPT into Change Agent.

Open ChatGPT → Settings → Data Controls
Click Export data
Download the ZIP file when emailed (note: It varies how long this takes. For some people it takes minutes; for some it takes as long as 24 hours)

Migrate to Change Agent

Step 3: Set Up Your Change Agent Account

Or, explore options for organization-wide Change Agent accounts and enterprise instances.

Step 4: Import Your Chat History (Optional)

After exporting your data from ChatGPT, you can import your full chat history into Change Agent. Note that only text will be ported over—images from your ChatGPT conversations will not transfer.

To Import Your Chat History:

Unzip the Export File: Locate the ZIP file you received via email from ChatGPT and extract its contents
Find Your Conversations File: Inside the unzipped folder, locate "conversations.json"—this contains all your chat history
Upload to Change Agent:
- Go to Settings > Data Controls > Import Chats
- Upload "conversations.json" directly from within your workspace settings

Now you’ll have access to all of your past chat history. You can even ask Change Agent to search your chat history to pull out info. For example, ask Change Agent “Search my chat history and summarize for me the tasks I created for my ChatGPT migration project.”

Step 5: Activate Change Agent Memory (Optional)

Turn on Memory in Change Agent

By default, Change Agent does not remember facts about you from chat to chat. But if you want a ChatGPT-like experience with Change Agent knowing your preferences and context, you can turn on Memory in the Personalization section of Settings:

Click the icon with your name at the bottom left of the screen, then open “Settings”
Find “Personalization” and click that
You will see “Memory.” On the right side of the window, press the small toggle button to turn this on (it will turn green).

Upload your ChatGPT personalization information into Change Agent

Now is the time to upload the bullet point information you collected in Step 1.

Open a new Chat in Change Agent. Type /import-ai-personalization to pull up the personalization prompt. Be sure to copy and paste your personalization bullet points into the message below where it says “<MEMORY BULLET POINT LIST HERE>”. You can edit those bullet point memories if you’d like.

Next you’ll see a long list of “Add Memory” operations. These should only take a couple minutes to complete.

Now, whenever you speak with Change Agent, it will reference these memories to personalize your experience to you, your organization, and the mission you care about. Change Agent will also keep these memories up to date as you chat with it.

Step 6: Add Knowledge & Create Workspace Apps

The key Change Agent features you’ll use to replicate your unique information and preferences from ChatGPT are Knowledge and Apps. Knowledge collections are repositories of documents you upload (e.g. campaign plans, strategy docs, handbooks) that can be referenced by Change Agent. Apps are reusable saved workflows you can use for a specific task or persona. You can click on the links to learn more about those features.

Activate Tools in Your Apps

Many Apps benefit from Change Agent's specialized tools for advocacy work, something not available within ChatGPT. To start using Tools:

Enable tools by default within Apps:
- Inside your App settings, toggle tools to be on by default like Action Network, ActBlue, LegiScan, or more
- This is not strictly necessary as you can always turn on tools within individual chat windows, but turning it on within your app means it will always be on in new chats
OR, configure tools within the chat window:
- Many tools require API keys to be set within the chat window
- Refer to the Change Agent Zendesk Tools article for setup help

Quick Reference: ChatGPT → Change Agent

ChatGPT Feature	Change Agent Equivalent
Custom GPTs	Workspace Apps
GPT Apps, Store, Plugins	Change Agent “Tools”
File uploads	Knowledge Collections
ChatGPT Personalization	Change Agent Personalization (Memory must be turned on)

Need Help?

If you need further assistance please contact us at support@thechange.ai, sign up for live support during our twice weekly office hours, or sign up for an AI Bootcamp cohort with other change makers like you.