GPTs as Force Multipliers: How Custom AI Can Augment and Enhance Your Nonprofit Productivity

This essay explores the transformative potential of integrating custom GPTs into nonprofit workflows.

Recently, I’ve started thinking about how GenAI platforms –and custom GPTs-–are high-level automation tools. While traditional automation streamlines routine tasks, custom GPTs can automate complex cognitive processes. These AI-driven tools do tasks faster. Also, they redefine what tasks can be automated, from donor communications to analysis of large data sets. In the landscape of nonprofit operations, custom Generative Pre-trained Transformers (GPTs) have great potential. By marrying the speed and reliability of automation with the fluency and capacity of Large Language Models (LLM), custom GPTs could shift how nonprofits operate–enabling even small organizations to magnify their impact.

Building GPTs with Open AI, or with open source tools, demonstrates this potential. OpenAI conceptualizes GPTs as customizable bots that use specific instructions to operate and filter with additional uploads of information. But, at their core, these GPTS work like highly specialized automations.

How does it work?

Every custom GPT is created with a set of specific instructions, or prompts, that automate specific tasks or routines. No matter what the topic, the process is the same. First, a user requests a solution (issues a prompt), then GPT searches the knowledge bank for relevant context and instructions to use to fulfill the request. Next, the GPT generates a response, checking the task with the larger LLM (such as ChatGPT-4) to ultimately generate a result (such as editing a document or drafting a letter).

I’ve built more than ten custom GPTs, all focused on specific tasks, and have become interested in how they create logic and structure to repeat a task, The process for how a GPT does this is the same every time.

1. A GPT uses its knowledge upload to retrieve information related to the request.
2. If there is a particular process, format, or style, data from the uploads are retrieved, reviewed, and sent to the LLM as information filters.
3. The GPT’s LLM then creates an internal text prompt based on the query as seen through the filter.
4. The LMM then uses this data to generate the requested output for the GPT.

When you create a custom GPT, you are building an automated template that incorporates new shared information from a user, like a prompt request, into its processes to create an output. When you use a custom GPT, whether it’s one you built yourself, or found in the GPT Store, you’re working with a GenAI automated process that filters the large language model (LLM) through a specific set of instructions (prompt directions) and filters (Knowledge uploads).

In other words, like a well-designed automated process, custom GPT models develop their capabilities by repeatedly performing tasks and recognizing patterns in vast datasets. The more data they process, the better they codify the “rules” and likely outputs for different inputs.

The implications of this understanding are powerful: If GenAI and GPTs can excel at automated processes, why not intentionally build them for repetitive tasks, such as editing writing in a particular voice or perspective, generating reports and letters, and summarizing notes? The most effective automation–and the most useful GPTs– augment people rather than replace them. This means that as we use GPTs to assist us, our roles will shift from being the sole author or editor to prompting, supervising, editing, and managing the GPt’s output.

My favorite custom GPT that I’ve built mirrors my writing voice and acts as an editor; my next favorite is a policy and data synthesizer for an education nonprofit that excels at reading, summarizing, and analyzing reports from the nonprofit’s perspective.

The automation of tasks such as editing, report generation, and donor communication through custom GPTs, taps into a broader trend that McKinsey & Company highlights. This trend is not confined to the nonprofit sector but is part of a larger shift towards generative AI as a driver in the next productivity frontier. The firm’s research underscores the potential of generative AI to automate activities that involve communication, documentation, and human interaction. Their report suggests a significant transformation in work dynamics across various fields, including education and technology. This transformation is poised to unfold sooner than previously anticipated, propelled by the capabilities of generative AI.

“…Many of the work activities that involve communication, supervision, documentation, and interacting with people, in general, have the potential to be automated by generative AI, accelerating the transformation of work in occupations such as education and technology, for which automation potential was previously expected to emerge later .”–The economic potential of generative AI: The next productivity frontier, McKinsey & Company,, June 14, 2023 | Report

To spell this out more directly, nonprofit staffers–and others in many fields–will see their need to transcribe, edit, and summarize meeting notes, compile individual biographies into a sheet for a grant application, or draft a case for support–become processes they can manage–and streamline–using Generative AI.

Custom GPTs offer knowledge workers a practical way to enhance our productivity and free up valuable time for higher-level work. Of course, to do this well, users will need to have some understanding of the logic of how an LLM works and how a GPT is constructed. They will also need to understand how the information provided in both the knowledge upload and the prompt for the custom GPT is organized, and whether rephrasing their prompt or asking it differently might lead to a better result.

For nonprofit leaders and teams ready to explore this area, the first step is to understand the capabilities of the AI tools and identify the areas where they can make the most significant impact. To learn more, engage with the growing community of AI experts and nonprofit professionals exploring this space. Try building some custom GPTs or work with someone who can do that with you.

The potential of using custom GPTs to automate and streamline knowledge work is not about replacing people. It’s about enhancing our work with tools that will allow us to do more.

Sidebar: Maximizing Impact with Custom GPTs in Nonprofits

Custom GPTs can be instrumental tools for nonprofits, streamlining high-value tasks and freeing up staff for more strategic activities. Here are examples of how they can make a difference:

• Fundraising Support: A GPT designed for fundraising can shift how donation appeals are crafted. By integrating donor profiles and past contributions, such a tool can produce highly personalized appeal letters. In practice, a task that might take several hours can be condensed.
• Meeting Efficiency: Transform the way meeting outcomes are recorded with a GPT that specializes in distilling minutes into actionable insights. Automatically transcribing discussions and highlighting decisions and action items with a trained GPT can save hours of manual transcription and summarization.
• Content Creation: A GPT tailored for content generation can be a game-changer. By analyzing existing reports, and publications, and writing with the organization’s unique voice, a Content GPT can draft website content, social media updates, and newsletters from notes and prompts.

By strategically integrating custom GPTs into their workflows, nonprofits can not only achieve significant time savings but also enhance the quality and personalization of their work.

If you’d like to get started in this area, get a subscription to ChatGPT-4, and check out my Guide to Building a Custom GPT. Or, get in touch with susan@collectiveagencyllc.com to talk about AI coaching, workshops, and support.