Good Vibes Bad Vibes: When to consider Vibe vs. Traditional Coding

AI is a reality in the modern world. On the surface, it gives users more power than ever in human history. While that is true, the LLMs (large language models) that you know, like ChatGPT, Claude, and Gemini, are double edged swords. 

The risks of AI are usually not nearly as dramatic as you hear in the news. It’s far stealthier. As intelligent as AI chat tools can sound, they are prone to misunderstandings in both directions. They often over assume what you tell them, and can leave out critical information when explaining to you what is being built. Here’s an example of what could happen in both cases.

A fundraising team is creating an app for donors to view metrics and make donations easily. On the front page, they want a big number with the dollar amount of all donations made in the last thirty days. When they write, “Show the amount donated this month”, the AI tool thinks they mean this calendar month, but never tells you that. Its assumption made the data wrong by reducing the time period the data comes from, and the reason why is hidden.

In the same app, the team is working on the donation payment page. They tell the AI, “Make the payment button”. The button appears on screen, and you can click it, so it all looks good. What the AI didn’t tell them is that the button routes to the same campaign every time, no matter the donor. Now there’s donations flowing into one campaign, and nothing into the campaigns the donors actually want. Worse, there’s no way to know until after the donations have already come in.

Keeping in mind issues can come up without knowing exactly how a vibe coding app works, it's helpful to think about some of the reasons you would consider vibe coding. The six examples below are reasons that could arise for any nonprofit, and will give you an idea of what to lean on and when to take a step back.

We have no in house programmer: Be cautious. The most common and obvious situation, especially for nonprofits, is because there isn’t someone who knows how to code available. This is especially tempting when there isn’t the budget to hire anyone else. While this might be the most needed example, it’s also the riskiest. Vibe coding is only productive when it can be tested by someone with the experience to spot the flaws. Some of this can be done by someone without technical experience, but only with significant time investment that may outweigh the benefits.

Copy of a popular app: Be cautious. Imagine an organization that manages subscription plans for donations to a variety of other nonprofits. They decide to create a social media app for their subscribers, to give them a community and inspire one another to give to their most favored causes. A team member has an idea to model it off Instagram. This is a huge challenge for the most experienced of developers. Successful, widely used Apps out there have a huge unseen library of algorithms, A/B testing, quality assurance and more that would be near impossible to duplicate by vibe coding. Rather than convince you of the difficulty of the task, an AI chat is more likely to run with your ideas, which could lead to disaster.

One off visual: Worth considering. You have a spreadsheet with donor giving and demographic data. There’s a meeting with leadership this afternoon, and they want a page on which you can click around and slice up the data. Ideally, you’d invest in a data visualization tool to make this replicable. If that isn’t needed, there’s less risk of misrepresenting the data. Vibe coding platforms like Base44 or Lovable are typically great for this, as they can take the spreadsheet, the requirement, and let you tell them what shape you want the buttons to be. Before you sink time into a vibe coded solution, just be sure you don’t need a simpler solution. For example, you could make a presentation with a tool like Gamma. Just because you can make an interactive dashboard doesn’t mean you should.

Mockup or prototype: Worth considering. A bunch of stakeholders expressed interest in an app that gives them key metrics and a form to submit leads in a single app. You have a bunch of ideas regarding the color scheme, how to organize the tabs, and how to label the data. Building an entire app for long term use is a bigger decision. A mockup is an easy decision. Vibe coding is far more efficient for human level testing like this. You can get what you need fast and get feedback before even considering next steps.

A unique app, just for your organization: Worth considering… cautiously. Every organization’s needs are different, and require a solution that wouldn’t work anywhere else. The level of customization you could get with vibe coding is particularly accessible, and far cheaper than a novel ground up project would run you with traditional programming. If the app is only being used by your own team, issues are more likely to be caught and cause fewer issues.

Testing your ideas and learning: the best reason! No matter your use case, vibe coding is a wonderful tool for exploration. One of the best features of using an AI chat is you can ask the model questions about what you want to do. They can all give you ideas, explain limitations, and do quick visual examples. After a bit of time tinkering with an app or dashboard, you’ll learn more about not only your own goals, but how to best write prompts that get where you want to go. Whether you end up doing it all in house, or use your experimentation as a mockup to get one of your own developers a head start, vibe coding will kick start your project.

Vibe coding has near unlimited use cases, and deserves your consideration. This is especially true if you’re able to put a little money towards tools like Base44 or Lovable. Keep in mind who’s working on the project, why you’re considering vibe coding, and what the goal is. If you think it’s worth a shot, then get started tinkering. The more you practice, the better you can leverage this new and powerful tactic to empower your nonprofit. Hopefully without breaking anything!