A prompt is the instructions you give Change Agent to complete any task.
As you use the product, you'll discover which inputs deliver better outputs. This guide provides best practices for prompting using the PROMPT framework, along with tips for getting started in the tool.
The Framework
P — Purpose & Positioning
Clearly state why you're asking and what strategic position you're taking. What change are you trying to create?
R — Role & Responsibility
Define who Change Agent should act as. Specify expertise, perspective, or organizational affiliation.
O — Output Action(s)
State what specific action or deliverable you need. Should Change Agent analyze, create, summarize, or something else?
M — Medium & Mechanics
Define how the output should be delivered. Specify length, format, style guidelines.
P — People & Power Audience
Identify who will receive this output. What are their values, concerns, and potential resistance points?
T — Truth & Transparency Guardrails
Establish what facts must be verified and what limitations must be acknowledged.
Sample PROMPT
P: We're raising $50K to fund a free legal clinic for undocumented immigrants, a direct response to rising deportations and lack of access to justice.
R: Act as a donor relations officer with 10 years in immigrant rights fundraising. Use warmth, urgency, and humility.
O: Draft a 200-word email that highlights 1 client story and 1 measurable outcome.
M: Format: 3 short paragraphs. Bold the client's name. End with a clear CTA: 'Donate now to protect [Client Name] and 100+ others.'
P: Audience: High-net-worth donors who care about immigration justice, emphasize impact, not guilt.
T: Never exaggerate outcomes or invent people's stories. Instead, say: 'I'll share a real story once we have permission or [insert immigrant story here].'
Step 1: Write Your Prompt + Add Context
Once you've defined your PROMPT framework, attach relevant documents, links, and examples that support each element. This "context" grounds your strategic positioning and ensures factual accuracy.
In Change Agent, the + Button can add:
- Links
- Documents from your computer
- Knowledge Collections saved in the Workspace
- Notes
- Past chats
- Screenshots
Tip: Tell Change Agent what you're uploading. For example: "I'm attaching our 2024 impact report—reference the volunteer statistics there"
Learn more about creating Knowledge Collections: Organizing Your Work with Knowledge Collections and Folders
Step 2: Click Send Message
After writing your prompt and attaching context, you must click Send Message for Change Agent to process your request. The AI won't respond until you submit.
Step 3: Edit and Refine (Prompt Iteration)
Review your output and request changes. Affirm what Change Agent got right and should keep, then make suggestions to iterate on your results.
Tip: The Regenerate button lets you test variations of your output without rewriting your entire prompt. Click it to see alternative versions, then choose the best one or combine elements.
Example prompts:
"This is a 9/10 urgency for tone, dial it back to 6/10"
"Make this half the length"
"Add data from our 2024 impact report" (then attach)
Step 4: Verify Facts & Watch for Hallucinations
A hallucination is when AI generates output that is factually incorrect, logically inconsistent, or entirely fabricated, while sounding plausible and confident.
How to Avoid Hallucinations:
- Provide context (upload docs)
- Use detailed prompting
- Break tasks into smaller steps
- Use the web search tool for current info
- Always verify critical facts and links
- Check what tools are on/off
- Open a new chat for new tasks
Multi-Turn Conversations
Complex work may require multiple exchanges. You can build context progressively across turns (e.g., "Search for recent news" → "Now synthesize the top 5 articles" → "Draft talking points"). Change Agent also now includes an optional Memory System that can store key information you share and recalls relevant details in future conversations.
Learn more about Memory and Universal Search: Memory and Universal Search
Privacy & Temporary Chats
For Sensitive Data (PII):
- Use Temporary Chats for donor names, addresses, or payment info
- Delete Immediately after use, especially for sensitive data
- Change Agent never sells, shares, or trains on your data
Tool Integrations
Some more advanced prompts may want to call on additional tools like LegiScan (legislative tracking), ActBlue (donor insights), Action Network (petition/event stats), or Gamma (slide deck generation). These require one-time API key setup in Integrations > Tools > Valves.
Learn more about Tools and Integrations: Tools and Integrations
Alternate Method: Ask Change Agent to Draft an Excellent Prompt
Don't want to build a prompt from scratch? You can ask Change Agent to write the prompt for you.
Sample Prompts:
"Draft an excellent prompt for summarizing a legislative bill for coalition partners"
"Make an excellent prompt for writing a donor thank-you note"
"Create a prompt for analyzing YouTube debate transcripts"
"Build a prompt for transforming our annual report into social media content"
Be sure to open up a new chat and paste the prompt in there to use it.
Prompting Pitfalls
- Not enough context: The more context you provide, the better outputs you’ll receive. If you do not provide enough context or relevant information, Change Agent may generate outputs that don't meet your needs.
- Poorly defined goals: Unclear or vague goals can result in outputs that aren't focused or effective.
- Inconsistent tone: Inconsistent tone can make the output seem disjointed or unprofessional.
- Overly broad prompts: Prompts that are too broad can lead to outputs that lack focus or clarity.
- Lack of specificity: Failing to provide specific details, such as target audience, tone, format, or key messaging, can result in outputs that aren't tailored to your needs.
- Ignoring feedback: Not taking the time to review and refine outputs based on feedback can lead to subpar results.
- Over-reliance on automation: Relying too heavily on Change Agent's automation features without reviewing and refining outputs can result in errors or inaccuracies.
- Overly complex prompts: Using overly complex or convoluted language in your prompts can confuse the model and lead to subpar outputs.
- Unclear pronouns: LLMs struggle to understand who pronouns refer to. It is better to repeat proper names and avoid he/she/it/they.
- Too many steps: LLMs aren’t yet great at breaking up multistep tasks. Try being very specific about each action you need to do in sequence or try breaking it up with multiple prompts.
Need Help?
If you need further assistance please contact us at support@thechange.ai, sign up for live support during our regularly scheduled office hours on Tuesdays and Thursdays, or sign up for an AI Bootcamp cohort with other change makers like you.