Building Custom GPTs for Your Nonprofit

A Custom GPT is a personalized version of ChatGPT that has been pre-loaded with your organization's voice, documents, and instructions. Once built, anyone on your team can use it to produce consistent, on-brand work without rewriting the same setup prompt every time. Custom GPTs are the single biggest leap in productivity a nonprofit manager can make after learning basic prompting.

What You'll Learn

• What a Custom GPT is and when it is worth building
• Step-by-step instructions to build your first Custom GPT for nonprofit work
• Three high-value Custom GPTs every nonprofit should consider
• How to share Custom GPTs with your team safely and effectively

What Is a Custom GPT?

A Custom GPT is a saved configuration of ChatGPT with:

• A set of instructions (the "system prompt") that defines its role, voice, and behavior
• A set of knowledge files you upload (your style guide, prior grants, brand voice doc)
• Optional web browsing, image generation, and code execution toggles
• A name, description, and icon

To build or use Custom GPTs, you need a ChatGPT Plus, Team, or Enterprise plan (as of April 2026, starting at $20/month). Free-tier users can use GPTs built by others but cannot build their own.

When to Build a Custom GPT

A Custom GPT pays off when a task is:

• Repetitive — you write donor thank-yous, grant LOIs, or board summaries regularly
• Voice-sensitive — you want consistent language across staff
• Context-heavy — you reuse the same background materials (mission, metrics, style guide)
• Multi-person — your team needs the same quality, not just you

If you only do a task twice a year, a Custom GPT is overkill. A saved prompt in a Google Doc works fine.

Step-by-Step: Building Your First Custom GPT

Step 1: Go to the GPT Builder

In ChatGPT, click your profile, then "My GPTs," then "Create a GPT." You will see a Configure tab — that is where the real setup happens.

Step 2: Define the Core Role

Give your GPT a name (e.g., "Grant Draft Assistant for {Org Name}"), a short description, and a clear instruction block. Example:

You are a senior grant writer for {Org Name}, a 501(c)(3) nonprofit that {mission}. Our primary audiences are family foundations, federal agencies, and corporate funders in {geography}. Your job is to help staff draft high-quality grant materials — LOIs, needs statements, program descriptions, and logic models — that align with our mission and voice. Always ask clarifying questions when the user's request is ambiguous. Never fabricate statistics; ask for source data instead. Write in a warm but professional voice. Avoid jargon like "underserved" or "at-risk" unless quoting a funder's own language.

Step 3: Upload Knowledge Files

Upload 5–10 reference documents:

• Your organization's one-pager or mission doc
• 2–3 prior successful grant proposals
• A style guide if you have one, or a list of voice do's and don'ts
• A list of program metrics and outcomes
• A glossary of your organization's key terms

The GPT will reference these automatically when drafting.

Step 4: Add Capabilities

Turn on Web Browsing if you want the GPT to research funders. Turn on Image Generation if you want it to produce appeal visuals. Turn off Code Interpreter unless you need data analysis.

Step 5: Test It

Ask it to draft a needs statement, an LOI opener, a thank-you letter. Refine the instructions based on what you like and do not like. Expect 3–5 iterations before the GPT produces consistently strong output.

Step 6: Share With Your Team

GPTs can be shared with anyone on your ChatGPT Team or Enterprise plan, or made public. Share within your team and train each person on how to use it.

Three High-Value Custom GPTs for Nonprofits

1. The Grant Draft Assistant

Purpose: Draft LOIs, full proposals, and needs statements in your organization's voice.

Knowledge to upload: prior winning proposals, mission doc, program descriptions, current strategic plan.

Instructions excerpt: Always ask for (a) the funder, (b) the amount, (c) any specific evaluation criteria before drafting.

2. The Donor Communications Assistant

Purpose: Thank-you letters, stewardship emails, appeal drafts, major donor meeting prep.

Knowledge to upload: your style guide, 5 examples of thank-yous and appeal emails you loved, the list of current programs, your voice guide.

Instructions excerpt: Always personalize using any notes provided. Never invent donor history or fabricate giving amounts. Ask for the donor's interests if not provided.

3. The Board Reporting Assistant

Purpose: Turn raw staff updates and financial data into clean board materials.

Knowledge to upload: prior board reports, your strategic plan, program dashboards, finance summary format.

Instructions excerpt: Always produce a 1-page executive summary with the sections Headline, Wins, Challenges, Financials, Decisions Required. Use plain language.

A Fourth GPT Worth Considering: The Impact Storyteller

Purpose: Turn beneficiary interview transcripts into case studies and appeal-ready narratives.

Knowledge to upload: your guidelines on ethical storytelling, 5 strong case study examples, a program glossary.

Instructions excerpt: Use only details present in the transcript. Never invent quotes or details. Always label composite or illustrative examples clearly.

Sharing and Governance

Before sharing a Custom GPT across your staff, consider:

• Access level. Keep GPTs with sensitive knowledge (donor lists, HR content) to staff only, never public.
• Data settings. On ChatGPT Team and Enterprise plans, data is not used to train models by default. Check your plan's data settings.
• Staff training. Schedule a 30-minute demo for anyone who will use the GPT. Show them how to prompt it, how to refine answers, and what to double-check.

When a Custom GPT Is the Wrong Tool

• If your nonprofit uses Google Workspace primarily, a shared Gemini Gem or NotebookLM may be simpler.
• If you need deep integration with your CRM, consider a workflow tool (covered in the next lesson) instead.
• If your use case is one-off, a saved prompt in a Google Doc is faster.

Worked Example

A mid-size environmental nonprofit built three GPTs over a weekend: a grant draft assistant, a donor comms assistant, and a press release assistant. They shared these across an 8-person team.

Six months later:

• Time-to-first-draft on grants dropped from 6 hours to 2
• Donor thank-you turnaround went from 7 days to 48 hours
• Press release drafts were produced in 20 minutes vs 90
• Voice consistency across staff improved measurably — the communications director reported a noticeable reduction in editorial rework

Key Takeaways

• A Custom GPT is a saved version of ChatGPT loaded with your instructions, voice, and reference documents
• They pay off for repetitive, voice-sensitive, context-heavy work shared across multiple staff
• Three starter GPTs for most nonprofits: Grant Draft, Donor Comms, Board Reporting
• Always upload reference documents (prior grants, style guides, program specs) for consistent output
• Keep governance tight: appropriate access, data settings, and staff training before wide rollout