A multi-year grant was an excellent fit for my client’s organization. The awardees would receive five years of generous unrestricted operating support, capacity-building coaching, and participation in a cohort of other grantees. How could Generative AI, specifically GPT (Generative Pre-trained Transformer) models, help me deliver a great proposal?

Introduction to GPT

Many nonprofit leaders are concerned about the fallout from having AI write grant applications for them. In my experience, no AI writes well enough that I’d trust it as a grant writer. Still, I have found AI helpful in analyzing data, summarizing information, and offering editorial suggestions based on specific guidance. When I had a chance to work on a large and complex grant proposal, it felt like the perfect moment to experiment with building and using two different GPTs, focused AI tools, to help me prepare and edit the grant.  

Writing a chronicle of my experience could also help nonprofit leaders think about what AI tools can do for them. 

Since I worked with a personal subscription account to Chat GPT-4 and tools I built that I can only use, the chances of others breaking in and seeing this data are close to zero. In addition, because my GPTs work with documents that I upload—i.e., I create a context for them to work within, with information that I supply—the chances for hallucinations are lowered (though they can happen).

So, in the hope that it will be helpful to others, here’s an account of my grant writing journey.

Custom GPT Tools

GPT is a language model developed by OpenAI that uses deep learning to generate human-like text. It can be fine-tuned for specific tasks, such as answering questions, generating content, or providing analysis based on input data.

I created two custom GPT tools to help me plan, analyze, write, and edit the grant proposal:

  • The first, The Brilliant Organization Brain, I built months ago. This GPT is based on uploads of the organization’s work, such as impact statements, program descriptions, and mission and vision. I’ve often examined data, reports, and proposals through the lens of the organization’s programs, impact, budget, and mission. It seemed like it would help analyze the upcoming grant through the lens of their work.
  • The second, Funder Focus, which I built at the start of this project, is a GPT solely dedicated to this RFP and reflecting the foundation’s mission, vision, and goals in awarding and administering this grant program. It knows all the grant rules and requirements and can analyze materials and assess strengths and weaknesses. For this GPT, I uploaded the application, announcement, FAQ, and information about five early grantees from the Foundation’s website.

While creating these custom tools required some technical expertise, the process involved fine-tuning pre-existing GPT models with relevant data and defining clear objectives for each tool. The development costs were minimal compared to the potential benefits, as the tools could be reused and adapted for future grant proposals. 

Since I have experience building GPTS (I’ve built about 15 so far), preparing and creating each one takes less than an hour. (If you’d like to make your own custom GPT–which I highly recommend–see my DIY how-to guide here. ( A related post on the power of using GPTs is here. )

Grant Analysis and Drafting Process

Preparing Materials

As always–and without any help from AI–I set up the preparation to write the grant. First, I created a folder with the grant materials. I started a Google Document with all the core information we would need: due date, submission portal, name of funder, all the directions, and a list of the questions we would need to answer. I also pulled together the core background materials we’d want to work with from the Brilliant Organization: board members, annual budgets, program budgets, impact reports, evaluations, program descriptions, and previous support requests, and uploaded them to the folder. 

I then read through the RFP, FAQ, information about a webinar, and everything on the website, downloaded and added everything that seemed helpful to the folder, and started working with the GPTs.

Analyzing Grant Requirements

First, using the Funder Focus GPT, I analyzed the grant materials, requirements, and previous winners to understand how to craft a competitive application. Specifically, I asked, “How can the Brilliant Organization make their application stand out?” The detailed response had some obvious answers (Aligned with foundation priorities) and more subtle ones (Demonstrated substantial, existing cross-sector collaboration).

After reading through and saving these notes, I used the Brilliant Organization Brain GPT to pull relevant data and language from the reports aligned with the grant’s focus and made notes.

Drafting Responses

Laying the questions out one by one with an anticipated word length, I started answering them.  However, I had a more conceptual ask: Since this was a big grant, wanting lots of systems change, what was The Brilliant Organization already doing that it could scale with these funds?   I turned to the  Funder Focus GPT for feedback since it was the RFP expert. 

Seeking GPT Feedback

The Funder Focus GPT suggested several critical program areas that could be emphasized to satisfy the grant’s expansion goals. After I reviewed the suggestions, I shared some of the questions and asked for tips on what to highlight to show how The Brilliant Organization initiative embodies the funder’s criteria. With that information, I worked on filling out the grant, using the Brilliant Organization GPT to help pull data from the uploads for my draft responses to the questions. 

As with any grant application, I filled in the first draft with core ideas, knowing we would iterate and revise as our understanding of the application evolved.

Revision and Editing

Incorporating Feedback

As my collaborators provided edits and feedback to the first draft, I used both GPT tools to refine the responses. The Brilliant Organization Brain helped me reflect on the organization’s perspective, and the Funder Focus GPT analyzed the strengths and weaknesses of each answer based on the grant criteria. I used both to help edit for clarity and word count.

Since I’d never used two GPTs for the same project before, it took me a little practice to figure out when to review and with which one, but having that second GPT focused on the grant’s goals was helpful. 

Final Review

It was the final rewrite time a few days before the application was due.  In a new Google document draft, I copied the questions and current answers, then pasted in all the comments and suggestions for each response. Within each question and answer, each reader or staffer’s input was uniquely coded with a different color for each question. This made tracking and incorporating feedback from various sources more linear, which seemed easier to analyze.  My job was now to assess all the input and pull a good revision together.

After reading everything shared for this grant proposal from start to finish three times, I decided to begin revising with the last question and work my way forward.  Based on the comments,  I rewrote a response for each question, then went back and forth between the two GPTs to check and refine my answer. I completed each response before moving to the next one.

Questions I asked myself as I worked included:

  • Was the answer compelling, accurate, and the right length?
  • What factual data and impact analysis could bolster our claims?
  • Did the response incorporate the feedback provided?
  • How could I make the writing vibrant and authentic?

Utilizing GPTs for Editing

I used the Brilliant Organization GPT to check and edit my revisions, with instructions such as: “Please edit the answer to this question. Shorten it to 20 words without losing the core meaning and supporting data, “and “Please review this long answer and tighten to 150 words. ”  

Then, I turned to the Funder Focus GPT to help check what I was preparing. One by one, I shared the  questions and responses in the chat box, with instructions like these: “Looking for an analysis of the strengths and weaknesses of this answer based on your uploads for the Funder Focus grant criteria.”

The  Funder Focus GPT reviewed my draft and responded with an in-depth analysis of the response. In the responses, I found that the Strengths were helpful, but the Weaknesses were invaluable. Comments like: 

  • Missing Direct Representation from Target Communities: 
  • Lack of Specificity on Missing Stakeholders
  • Potential Overemphasis on Institutional Partners

were followed by recommendations for improvement, such as

  • Enhance Community Representation:
  • Clarify Roles of Missing Stakeholders: 
  • Balance Institutional and Community Leadership:

As I went back and reviewed and revised, I continued to share modifications with the GPTs, both for feedback and to help in editing for word count.  I didn’t accept every suggestion, but it was helpful to see the options provided.

Fact-checking and Final Touches

Once I had a more finished revision, with the approximate correct word length for each answer,  I checked the statistics used in the proposal with the Brilliant Organization GPT. 

Specifically, I asked, “What accurate data can you pull from your knowledge upload for this answer? Provide the PDF source for each data point we mention.”  

Once I had this information, I double-checked it, using copies of the PDFs and the website. I wanted to be 100% confident that all reported data was accurate and had a known source. I then read, reviewed, and edited once more.

After completing my work, I turned the final draft over to the final reviewer to review and submit. 

Ethical Considerations

As I used these AI tools, these ethical considerations were top of mind:

  • Maintaining human authorship: While  I used AI tools to assist in the process, humans wrote and critiqued our proposal. It truly reflected the work and ideas of the people involved and the organization we represent.
  • Ensuring accuracy and accountability: I did not assume AI would know if it made an error. Building a GPT with our data and using that as a primary source helped minimize error. Still, we had to fact-check and verify the information generated by AI tools. It was our responsibility to ensure the accuracy and integrity of the submitted proposal.
  • Adhering to grant writing ethics: AI tools should be used consistently with established grant writing ethics and best practices. Using a GPT helped avoid plagiarism, but I was still alert for accidental misrepresentation or data manipulation.

Benefits of Using AI

Was this all more trouble than it was worth? Some people might think so.  However, this was a productive effort as I seek to become more proficient in AI.

Using AI tools in this grant writing process provided several benefits:

  • Time-saving: The GPT tools could synthesize and analyze large amounts of data at a rate and quantity that I could not match in the same time frame. This saved me valuable time in the drafting and revision process.
  • Insightful feedback: The AI-generated suggestions and analyses offered fresh perspectives and helped identify areas for improvement that I might have overlooked.

Challenges and Limitations

Nevertheless, it wasn’t all smooth sailing.  While the GPT tools were invaluable in saving time and providing helpful feedback, there were some challenges and limitations:

  • Ensuring accuracy: I double-checked all data and facts generated by the GPT tools to provide 100% accuracy and proper attribution. I found one area where the Brilliant Organization GPT quoted stats that were not in our uploads and removed them.
  • Maintaining authenticity: Preserving our organization’s authentic voice and style for this application was essential.
  • Overreliance on AI: AI is a tool to assist the process, not a complete solution. The AI helped in some areas, but human judgment and expertise drove the proposal.

Conclusion

Although I have been using AI to help with aspects of my writing, fundraising, and development work since 2023, this was the first time I used two custom GPTs to work on a large grant. These tools saved me time summarizing data and analyzing content.  They also provided helpful feedback and synthesis as I sought to align our proposal with core grant objectives in an accurate but competitive manner.

We recently submitted the grant proposal and are awaiting the results. Regardless of the outcome, I feel good about the quality of our submission and the work we put into it. Using custom GPT tools to assist in the grant writing process proved valuable, saving time and providing helpful insights while requiring human oversight and judgment.

If you have experience using AI in your grant writing process or have questions about this approach, I’d love to hear your thoughts and experiences in the comments below.

 

A multi-year grant was an excellent fit for my client’s organization. The awardees would receive five years of generous unrestricted operating support, capacity-building coaching, and participation in a cohort of other grantees. How could Generative AI, specifically GPT (Generative Pre-trained Transformer) models, help me deliver a great proposal?

Introduction to GPT

Many nonprofit leaders are concerned about the fallout from having AI write grant applications for them. In my experience, no AI writes well enough that I’d trust it as a grant writer. Still, I have found AI helpful in analyzing data, summarizing information, and offering editorial suggestions based on specific guidance. When I had a chance to work on a large and complex grant proposal, it felt like the perfect moment to experiment with building and using two different GPTs, focused AI tools, to help me prepare and edit the grant.  

Writing a chronicle of my experience could also help nonprofit leaders think about what AI tools can do for them. 

Since I worked with a personal subscription account to Chat GPT-4 and tools I built that I can only use, the chances of others breaking in and seeing this data are close to zero. In addition, because my GPTs work with documents that I upload—i.e., I create a context for them to work within, with information that I supply—the chances for hallucinations are lowered (though they can happen).

So, in the hope that it will be helpful to others, here’s an account of my grant writing journey.

Custom GPT Tools

GPT is a language model developed by OpenAI that uses deep learning to generate human-like text. It can be fine-tuned for specific tasks, such as answering questions, generating content, or providing analysis based on input data.

I created two custom GPT tools to help me plan, analyze, write, and edit the grant proposal:

  • The first, The Brilliant Organization Brain, I built months ago. This GPT is based on uploads of the organization’s work, such as impact statements, program descriptions, and mission and vision. I’ve often examined data, reports, and proposals through the lens of the organization’s programs, impact, budget, and mission. It seemed like it would help analyze the upcoming grant through the lens of their work.
  • The second, Funder Focus, which I built at the start of this project, is a GPT solely dedicated to this RFP and reflecting the foundation’s mission, vision, and goals in awarding and administering this grant program. It knows all the grant rules and requirements and can analyze materials and assess strengths and weaknesses. For this GPT, I uploaded the application, announcement, FAQ, and information about five early grantees from the Foundation’s website.

While creating these custom tools required some technical expertise, the process involved fine-tuning pre-existing GPT models with relevant data and defining clear objectives for each tool. The development costs were minimal compared to the potential benefits, as the tools could be reused and adapted for future grant proposals. 

Since I have experience building GPTS (I’ve built about 15 so far), preparing and creating each one takes less than an hour. (If you’d like to make your own custom GPT–which I highly recommend–see my DIY how-to guide here. ( A related post on the power of using GPTs is here. )

Grant Analysis and Drafting Process

Preparing Materials

As always–and without any help from AI–I set up the preparation to write the grant. First, I created a folder with the grant materials. I started a Google Document with all the core information we would need: due date, submission portal, name of funder, all the directions, and a list of the questions we would need to answer. I also pulled together the core background materials we’d want to work with from the Brilliant Organization: board members, annual budgets, program budgets, impact reports, evaluations, program descriptions, and previous support requests, and uploaded them to the folder. 

I then read through the RFP, FAQ, information about a webinar, and everything on the website, downloaded and added everything that seemed helpful to the folder, and started working with the GPTs.

Analyzing Grant Requirements

First, using the Funder Focus GPT, I analyzed the grant materials, requirements, and previous winners to understand how to craft a competitive application. Specifically, I asked, “How can the Brilliant Organization make their application stand out?” The detailed response had some obvious answers (Aligned with foundation priorities) and more subtle ones (Demonstrated substantial, existing cross-sector collaboration).

After reading through and saving these notes, I used the Brilliant Organization Brain GPT to pull relevant data and language from the reports aligned with the grant’s focus and made notes.

Drafting Responses

Laying the questions out one by one with an anticipated word length, I started answering them.  However, I had a more conceptual ask: Since this was a big grant, wanting lots of systems change, what was The Brilliant Organization already doing that it could scale with these funds?   I turned to the  Funder Focus GPT for feedback since it was the RFP expert. 

Seeking GPT Feedback

The Funder Focus GPT suggested several critical program areas that could be emphasized to satisfy the grant’s expansion goals. After I reviewed the suggestions, I shared some of the questions and asked for tips on what to highlight to show how The Brilliant Organization initiative embodies the funder’s criteria. With that information, I worked on filling out the grant, using the Brilliant Organization GPT to help pull data from the uploads for my draft responses to the questions. 

As with any grant application, I filled in the first draft with core ideas, knowing we would iterate and revise as our understanding of the application evolved.

Revision and Editing

Incorporating Feedback

As my collaborators provided edits and feedback to the first draft, I used both GPT tools to refine the responses. The Brilliant Organization Brain helped me reflect on the organization’s perspective, and the Funder Focus GPT analyzed the strengths and weaknesses of each answer based on the grant criteria. I used both to help edit for clarity and word count.

Since I’d never used two GPTs for the same project before, it took me a little practice to figure out when to review and with which one, but having that second GPT focused on the grant’s goals was helpful. 

Final Review

It was the final rewrite time a few days before the application was due.  In a new Google document draft, I copied the questions and current answers, then pasted in all the comments and suggestions for each response. Within each question and answer, each reader or staffer’s input was uniquely coded with a different color for each question. This made tracking and incorporating feedback from various sources more linear, which seemed easier to analyze.  My job was now to assess all the input and pull a good revision together.

After reading everything shared for this grant proposal from start to finish three times, I decided to begin revising with the last question and work my way forward.  Based on the comments,  I rewrote a response for each question, then went back and forth between the two GPTs to check and refine my answer. I completed each response before moving to the next one.

Questions I asked myself as I worked included:

  • Was the answer compelling, accurate, and the right length?
  • What factual data and impact analysis could bolster our claims?
  • Did the response incorporate the feedback provided?
  • How could I make the writing vibrant and authentic?

Utilizing GPTs for Editing

I used the Brilliant Organization GPT to check and edit my revisions, with instructions such as: “Please edit the answer to this question. Shorten it to 20 words without losing the core meaning and supporting data, “and “Please review this long answer and tighten to 150 words. ”  

Then, I turned to the Funder Focus GPT to help check what I was preparing. One by one, I shared the  questions and responses in the chat box, with instructions like these: “Looking for an analysis of the strengths and weaknesses of this answer based on your uploads for the Funder Focus grant criteria.”

The  Funder Focus GPT reviewed my draft and responded with an in-depth analysis of the response. In the responses, I found that the Strengths were helpful, but the Weaknesses were invaluable. Comments like: 

  • Missing Direct Representation from Target Communities: 
  • Lack of Specificity on Missing Stakeholders
  • Potential Overemphasis on Institutional Partners

were followed by recommendations for improvement, such as

  • Enhance Community Representation:
  • Clarify Roles of Missing Stakeholders: 
  • Balance Institutional and Community Leadership:

As I went back and reviewed and revised, I continued to share modifications with the GPTs, both for feedback and to help in editing for word count.  I didn’t accept every suggestion, but it was helpful to see the options provided.

Fact-checking and Final Touches

Once I had a more finished revision, with the approximate correct word length for each answer,  I checked the statistics used in the proposal with the Brilliant Organization GPT. 

Specifically, I asked, “What accurate data can you pull from your knowledge upload for this answer? Provide the PDF source for each data point we mention.”  

Once I had this information, I double-checked it, using copies of the PDFs and the website. I wanted to be 100% confident that all reported data was accurate and had a known source. I then read, reviewed, and edited once more.

After completing my work, I turned the final draft over to the final reviewer to review and submit. 

Ethical Considerations

As I used these AI tools, these ethical considerations were top of mind:

  • Maintaining human authorship: While  I used AI tools to assist in the process, humans wrote and critiqued our proposal. It truly reflected the work and ideas of the people involved and the organization we represent.
  • Ensuring accuracy and accountability: I did not assume AI would know if it made an error. Building a GPT with our data and using that as a primary source helped minimize error. Still, we had to fact-check and verify the information generated by AI tools. It was our responsibility to ensure the accuracy and integrity of the submitted proposal.
  • Adhering to grant writing ethics: AI tools should be used consistently with established grant writing ethics and best practices. Using a GPT helped avoid plagiarism, but I was still alert for accidental misrepresentation or data manipulation.

Benefits of Using AI

Was this all more trouble than it was worth? Some people might think so.  However, this was a productive effort as I seek to become more proficient in AI.

Using AI tools in this grant writing process provided several benefits:

  • Time-saving: The GPT tools could synthesize and analyze large amounts of data at a rate and quantity that I could not match in the same time frame. This saved me valuable time in the drafting and revision process.
  • Insightful feedback: The AI-generated suggestions and analyses offered fresh perspectives and helped identify areas for improvement that I might have overlooked.

Challenges and Limitations

Nevertheless, it wasn’t all smooth sailing.  While the GPT tools were invaluable in saving time and providing helpful feedback, there were some challenges and limitations:

  • Ensuring accuracy: I double-checked all data and facts generated by the GPT tools to provide 100% accuracy and proper attribution. I found one area where the Brilliant Organization GPT quoted stats that were not in our uploads and removed them.
  • Maintaining authenticity: Preserving our organization’s authentic voice and style for this application was essential.
  • Overreliance on AI: AI is a tool to assist the process, not a complete solution. The AI helped in some areas, but human judgment and expertise drove the proposal.

Conclusion

Although I have been using AI to help with aspects of my writing, fundraising, and development work since 2023, this was the first time I used two custom GPTs to work on a large grant. These tools saved me time summarizing data and analyzing content.  They also provided helpful feedback and synthesis as I sought to align our proposal with core grant objectives in an accurate but competitive manner.

We recently submitted the grant proposal and are awaiting the results. Regardless of the outcome, I feel good about the quality of our submission and the work we put into it. Using custom GPT tools to assist in the grant writing process proved valuable, saving time and providing helpful insights while requiring human oversight and judgment.

If you have experience using AI in your grant writing process or have questions about this approach, I’d love to hear your thoughts and experiences in the comments below.