New to AI? This guide for Chief Financial Officers and other senior finance leaders of nonprofits explains generative AI, top chatbot tools, and how they and their employees can start using AI with confidence.
Nonprofit finance teams can start using AI today. Generative AI tools such as ChatGPT, Claude, Copilot, and Gemini can summarize grant terms, explain variance drivers, and draft reports, but they predict rather than think, so everything they produce needs human review. Start with the problem you want to solve, keep confidential data out of free tools, and put governance in place before you scale.
Getting Started Using AI in Nonprofit Finance Departments
Everyone was a beginner at some point. That’s a key truth for learning anything. Generative AI (Artificial Intelligence) is no exception, which means now is a good time to get started. That was an undercurrent at many of the sessions at JMT Consulting’s Innovate 2026, including “From Buzzword to Basics: Understanding AI” presented by keynote speaker Ari Gunzburg and “The Future of Finance: AI in Accounting and the Workforce Transformation” presented by Neil Shah, founder of Altruva AI.
Both emphasized that, collectively, we’re much earlier in the AI adoption process than we might think, so you’re not behind if you’re just getting started with it. “You are really not too late to start an AI journey,” said Gunzburg. “And it’s only going to get stronger from here.”
The field of artificial intelligence, however, is not new. According to a history published by Stanford University, the term artificial intelligence was coined by computer scientist John McCarthy in a proposal he wrote to hold a workshop in 1956. And that workshop, named the Dartmouth Summer Research Project on Artificial Intelligence, is where the study of AI was “officially” started.
But its roots run even deeper. For instance, British mathematician Alan Turing published the paper “Computing Machinery and Intelligence” in 1950, in which he asked, “Can machines think?” And earlier, in 1943, mathematicians Warren McCulloch and Walter Pitts created the first mathematical model of a neural network, based on Boolean logic, which was invented around 1847 by George Boole, where there can only be a true and a false (i.e., 1 or a 0) and inspired by Turing’s 1936-7 “mathematical notion of computation.”
“Alan Turing said that, in the future, we will have this universal machine that could perform any calculation,” said Gunzburg during his presentation.
And we have been using simpler versions of artificial intelligence for years. Examples include anytime we use spell check or the suggestions for the next word while writing a text, email, or longer document or click on “you might also like” suggestions while shopping on Amazon and other online stores or streaming shows on such services as Netflix. Even your social media feed, spam filter, and search results are now informed by AI.
“Modern AI could be said to start with the launch of ChatGPT in 2022,” said Gunzburg. “Of course, OpenAI started before that, but that’s when they made it accessible to people with ChatGPT.” Additional generative AI tools soon followed.
Today’s more advanced AI tools can help CFOs in nonprofit organizations speed up their month- and year-end close-outs and financial analysis and reporting; improve forecasting and scenario planning capabilities; create accessible leadership and board presentations; and more. And by enhancing these capabilities, they will also help reduce the administrative burden on the finance team, which tends to run very lean in nonprofits.
Since generative AI has only been available to the public since 2022, it is no surprise if you are just getting started with it. While the fastest way to learn generative AI is to experiment with it, this approach can often be frustrating, because you might not get the results you’d like at first.
To help you get started while minimizing any frustration, here’s a brief overview of Artificial Intelligence (AI) and how it can help people working in finance for nonprofit organizations:
A Definition of Generative AI and Why You Should Care
Before we dive into the different generative AI tools you can use, we need to be clear on its definition. That’s because there are different types of AI. We’re just hearing about this one because it was such a powerful, transformative breakthrough.
Like all other types of AI, generative AI doesn’t “think” like us. Rather, it essentially just makes predictions. It simply draws from the information available to it to predict what should come next based on what was requested. In other words, it creates content “by learning patterns from large sets of sample material.”
“AI is not magic. It’s not sci-fi. It is just pattern recognition trained on a massive amount of data,” said Gunzburg. “I can give something to it with a little bit of context about what I want, and it will give me an answer.”
Generative AI can help CFOs with a wide variety of tasks, including:
- Summarize grant terms,
- Explain variance drivers,
- Draft emails, write reports and updates
- Develop summaries,
- Complete calculations,
- Create onboarding materials,
- Brainstorm policy checklists,
- And more.
It can even help you think through how to communicate difficult information and what to say.
We should keep in mind that generative AI tools have been programmed to be compliant and to always give an answer that sounds confident. The tool doesn’t “understand” what it is producing and its output can be biased or based on outdated data. “If there’s a bias on the internet about a certain thing, AI might carry that over,” said Gunzburg. “You have to go down the rabbit hole to make sure that you get to a reliable source, and it says what the AI reported it to say.”
What’s more, the content could be simply wrong because of problems with the input, such as the tool drawing from irrelevant or conflicting “sample material” or because the prompt was poorly worded.
That means generative AI should not be treated as an authority. Everything it produces needs to be scrutinized and refined by you. “AI is a tool, not a decision maker,” said Gunzburg. “You are still the expert. Own the outcome.”
The Top Generative AI Tools for Nonprofits Available Today
To get started, you might want to experiment with a few of the various tools. Here are some of the top generative AI tools:
- ChatGPT (by OpenAI): The one that started it all. While ChatGPT is good for general-purpose brainstorming, you must keep in mind that the generated content will come across as confident even when it’s wrong or serving up hallucinations, so you’ll need to apply a critical eye to what is generated and confirm everything that it produces. To reduce the risk of hallucinations, ask for explicit citations in your prompt.
- Claude (by Anthropic): If you have complex instructions, this one might be the right one for you. It is also good for long-form content generation and for document analysis and summarization. However, this assistant has also been known to hallucinate and fabricate details, although it is better than ChatGPT at admitting when it is uncertain. Like above, to reduce this risk, request the explicit citations in your prompt.
- Copilot (by Microsoft): If your organization uses any Microsoft products, you probably already have access to the paid version of this tool. It is helpful for summarizing Teams meetings, drafting emails, creating slides, analyzing content from Excel, SharePoint, and other Microsoft applications, and more, although it has been known to hallucinate. The controls you have for Copilot will depend on the license you hold.
- DeepSeek: A low-cost alternative, DeepSeek has been banned for use on a government device by multiple countries because they “are concerned about the security risks posed by the Chinese application,” so it is not recommended for analyzing highly confidential enterprise, regulated, or government data. The risk of hallucinations is about the same as the assistants discussed above.
- Gemini (by Google): The content it generates can feel less polished than the tools described above, but because of its integration with the Google search engine, it is good for conducting deeper research. Like Copilot for Microsoft products, Gemini is integrated with the other tools in the Google ecosystem and works seamlessly with Google Gmail, Docs, Sheets, Slides, Meet, Drive, and more.
- Perplexity: An AI-based search-and-answer engine: “Unlike traditional AI models that generate responses from internal data, Perplexity focuses on … providing real-time, cited answers.” As a result, there’s a lower risk that results will include hallucinations. However, you should still check these citations to confirm they are accurate.
“One fun thing to do is to open a few different AIs and give each the same prompt and then read through the output from all of them,” said Gunzburg. “When you play them against each other, you start to find the gaps.”
You might also want to check out Canva for design; Grammarly for writing help; ElevenLabs for voice generation, Runway for video generation, Suno for generating music and songs based on your prompts, and other AI tools for generating other forms of content.
To get started, many of these have free versions, but keep in mind that you are “the product” when you are not paying for it. In other words, they could aggregate and sell your data and/or co-opt your proprietary data for training purposes. That means you should never place health information, personally identifiable information, strategic plans, and other confidential materials on a free unmanaged version and even on some of the paid tiers or plans.
The Key Types of Artificial Intelligence Beyond Generative
While generative AI seems to be getting all the buzz in the popular media, it’s not the only type of AI available today. The top four other functional types are:
- Agentic AI: These are “AI systems that are semi- or fully autonomous.” They combine “with other software systems to complete tasks independently or with minimal human supervision.” CFOs in nonprofits can use Agentic AI for multistep processes like expense reconciliation, grant compliance tracking, and more.
- Automation and Workflow AI: This simpler type of AI helps you “optimize business processes and reduce human intervention in routine operations,” cutting “time spent on repetitive tasks.” Use it to match purchase orders to invoices; process recurring invoices; flag unusual transactions; and more.
- Conversational AI: This is “a type of AI that can simulate human conversation. It is made possible by natural language processing (NLP),” enabling you to locate data, take notes, and complete other computer-based tasks simply by talking in everyday language to a voice-enabled computer.
- Predictive AI and Analytics: Probably not the best category name since all AI works by making predictions, this type of AI combines “statistical analysis with machine learning algorithms to find data patterns and forecast future outcomes. It extracts insights from historical data to make accurate predictions about the most likely upcoming event, result or trend.” Predictive AI can help CFOs in nonprofits forecast future cash flow and more. But there’s one caveat: For an accurate prediction, the model design and assumptions must be accurate, and the data must be clean.
Of course, there are probably other ways to “slice-and-dice” the AI pie. But this list should give you a sense of the depth of AI capabilities available to you and your team. And because of this wide variance, the place to start is not with the type of AI, but with the problem you are trying to solve or goal you’re aiming to achieve.
The Rewards and Risks of Using an AI Assistant at Work
In addition to all the benefits listed earlier, introducing AI tools to the finance department in a nonprofit can help the team:
- Improve the quality of their “what-if” scenario planning;
- Develop more accessible financial communications for boards and program leaders
- Enable stronger anomaly detection and other controls; and
- Reduce the administrative burden on the finance teams, among other improvements.
And it gives CFOs the time and the tools to go from being a “scorekeeper,” as Shah referred to it in his presentation, to becoming a more forward-looking “strategic architect.”
This matters for nonprofits because their finance departments are often overburdened. As we all know, many if not most finance teams in nonprofits run lean for what they are asked to produce, offering them little time to be strategic.
But if used carelessly, AI tools can enable confidentiality breaches, compliance problems, and other issues resulting in reputational harm or worse. That’s because generative AI has been known to invent facts, “misread” documents, produce incorrect calculations, create fake citations, overgeneralize, provide outdated information, “confuse” similar concepts and more.
For nonprofit CFOs, this risk is significant. Just imagine what these errors can create. It can be anything from making poor financial decisions to a loss of funding.
“AI can produce answers that sound confident but are incorrect,” Gunzburg reported. “The best way to think of AI is to imagine that it is an intern that has read every document ever written. But would you hand an intern the decision-making power? If we expect the AI to be like an executive and make those big decisions, it will eventually get us into trouble.”
The best way to keep yourself and your organization out of trouble? Simply stated, it’s selecting the right AI tools, governance policies, employee training, and how you position the AI-enabled tools within the finance department.
Selecting, Introducing, and Managing AI Accounting Tools
There are way too many steps and considerations for implementing AI into your nonprofit’s financial systems to list them in this article, but here are the most important points to start your thinking.
The first thing to keep in mind when considering any type of AI tool is what is known as “agent washing,” which is when technology providers rebrand their ERP solutions that have more basic automation capabilities and simple AI chatbots and promote them as being advanced AI agents, overstating the role that artificial intelligence plays in them.
Instead of getting caught up in the “bells and whistles” of promised capabilities, focus, as stated earlier, on achieving your goals and on addressing your business needs. That focus will help guide you into determining the right AI tools for your organization, even if they are not the most advanced. Gunzburg suggested that your first question should be, “Can it do what I want it to do?” Then, as he stated, you should ask “What data was it trained on? Is it generic data or is it on ERP data that applies to what I’m doing? Does it show its work, so that I can check it? Is the output factually correct? Can you override it or does it make decisions for you?”
Beyond business fit, you’ll then need to explore how it handles data and privacy, security (such as sign on controls and breach notification processes), contract terms, and more.
Next, think about how you’re going to position these new tools with your finance team when you introduce them. Many employers are positioning their advanced AI tools as co-workers, but this is proving to be problematic.
In research published by Harvard Business Review, the researchers stated that “in a randomized experiment, we found that humanizing AI can shift accountability away from individuals,” leading to errors slipping through in the review process. For instance, when tasked with reviewing the output of an “AI employee” framing versus an “AI tool” framing to find errors, “participants that reviewed documents with the AI employee framing caught 18% fewer errors.”
The bottom line: More mistakes and inaccuracies made it through the review process when the content was framed as coming from an “AI employee.”
The researchers also wrote that “When AI was framed as an employee rather than as a tool, personal accountability fell by 9 percentage points, while accountability attributed to the AI rose by 8 percentage points. This is problematic, because today’s AI systems cannot be held accountable and require clear human ownership.”
Another concern is that human employees will be replaced by AI. But as Gunzburg pointed out, “AI is not going to be replacing anybody. What’s going to be replaced are the people not willing to use AI.” In fact, it’s predicted that the increased productivity realized by AI might even end up increasing the overall number of jobs. It is just that these jobs will be different and will require retraining. In fact, according to Gartner, a business research and advisory company, “AI won’t cause a jobs apocalypse, but it will unleash job chaos.” Gartner found that starting in 2028-2029, AI will create more jobs than it eliminates, yet over 32 million jobs will be significantly transformed each year.
Said Gunzburg, “A lot of companies that laid off people are scrambling right now trying to figure out how to get those people back. AI is an extremely valuable tool, but it will not be taking away jobs at this moment.”
The bottom line: You should reassure your team that they’re not going to be replaced as well as provide the needed training to welcome them to the AI era.
Finally, and perhaps the most important consideration, you need to develop the guidelines and the governance processes. You need to decide on which AI tools to adopt, who can use them, what data can they enter, who will review the output, how will you manage risk, who will serve on the AI governance committee, and much more.
This part all might sound overwhelming, but if you’re with a nonprofit, we at JMT Consulting are here for you. We have helped numerous nonprofit organizations make the jump to AI-enabled financial tools. Our services include helping to strategically evaluate technology and modernize finance operations. We also use a human-centered approach to support your teams through the change. Whether your nonprofit is seeking to implement AI tools or modernize your finance processes and operations in other ways, contact us today.
Frequently Asked Questions
What is Generative Artificial Intelligence (AI)?
Generative AI is essentially a “prediction machine” that generates content (such as text, pictures, code, and more) in response to instructions from a user. It works by drawing from the large amount of information available to it to find patterns that enable it to predict what should come next based on what was requested. As a result, it generates the new content.
What is the best Generative AI Assistant to use if I want to get started?
It depends on your needs and expectations. For a good, general-purpose assistant, there is ChatGPT. If you use Microsoft products such as Microsoft 365, then start with Microsoft Copilot. Likewise, if you use Google products such as Google Docs and Sheets, then start with Gemini. Meanwhile, Perplexity is best if you want to conduct research on the web. And finally, Claude is best for generating longer form, more human sounding content as well as for analyzing and summarizing longer documents.
What are the various functional forms of Artificial Intelligence (AI)?
There are five different functional “types” of AI available today: 1) Agentic AI that works independently to figure out and complete complex tasks with little or no supervision by a human; 2) Automation and Workflow AI that automates any repetitive tasks, eliminating the need for a human being to complete them; 3) Conversational AI that enables a human to speak to a computer in their natural language and having the computer respond verbally in the same natural language with the requested information or action; 4) Generative AI that develops new content (see the FAQ above); and 5) Predictive AI and Analytics that is used for forecasting based on statistically analyzing the data it has.
Is it safe to put financial data into free AI tools?
Not if the data is confidential. When you use a free version, you are often “the product,” and your data could be aggregated, sold, or used for training. Never enter health information, personally identifiable information, strategic plans, or other confidential materials into a free, unmanaged tool, and check the data terms of paid plans too.
Will AI replace nonprofit finance staff?
Not at this moment. Gartner projects that starting in 2028-2029, AI will create more jobs than it eliminates, although more than 32 million jobs will be significantly transformed each year. As Gunzburg put it, the people who will be replaced are the ones not willing to use AI, so invest in training your team.
