WHAT YOU WILL LEARN
- Why Veracross spent a year building its AI layer before releasing a single product
- The seven principles that define how Veracross approaches AI for independent schools
- How VC Studio turns plain-language questions into reports, dashboards, and visualizations from your live school data
- Why clean, unified data architecture is the foundation of trustworthy AI outputs
Veracross Chief Architect Steve Erickson on the design principles behind VC Studio, Veracross’s AI-powered reporting tool, and why thoughtful engineering matters as much as being first to market.
School software can’t be approached like any other industry when it comes to AI. “Move fast and break things” can’t be a guiding principle when you’re responsible for student data, for family trust, and for decisions that affect children’s lives. When an AI-powered tool comes along, the right instinct isn’t just to get excited; it’s to ask whether anyone has thought it through.
We’ve been thinking it through. Veracross has been a data company since day one, and we’ve been a reporting company for 25 years, always trying to make it easier for schools to report on their data. When AI first came along, it was a natural instinct to ask: how can we use this to help schools? But we also knew we had to take the time to get it right.
Over the past year we’ve built what we call Veracross Intelligence: an AI layer that we will thoughtfully extend across the Veracross platform over time. VC Studio is the first product we’re launching on top of it. It’s an AI-powered data analytics tool that turns plain-language questions into finished outputs: detailed reports, interactive dashboards, and polished visualizations drawn directly from your school’s live data.
Building it required a clear set of commitments about what AI should and shouldn’t do in a school environment. Here are our seven core principles.
1. AI is a fundamental paradigm shift, not just a feature.
The internet, mobile phones, AI: these are the three most transformative technologies of the past 30 years. All software is eventually going to have to become AI-native, and that’s happening faster than people may have realized. We have to approach AI development with that level of seriousness.
2. AI is a tool to achieve valuable outcomes, not for its own sake.
We kept asking ourselves: what can we work on that only we can do? The answer was easy. We are good at helping schools manage their data chaos. Reporting is an evergreen need all institutions have, and it is hard. In some ways, this is the whole reason for Veracross’s existence. We’ve been a reporting company for 25 years. When AI came along, it was natural to ask: how can we use this to help schools report on their data?
3. Be more ambitious. Build beyond just chat and text.
We didn’t want our first AI foray to be for a single department or for a single task. We wanted something multi-department, handling a variety of use cases. And we built it to be bigger than the SIS. The underlying AI platform technology is set up and ready for us to fit for purpose. Reporting and visualizations are first, but the vision is broader.
4. AI should amplify people, not replace them.
Reporting has always forced a bad tradeoff: powerful tools like Tableau or Crystal Reports require certification training, while approachable tools like Excel or Google Looker limit what you can build. On top of that, you have the data pipeline problem of getting data out of your system and into the reporting tool. With VC Studio, we solve both.
Out of the box, we have a data pipeline and data lake already set up. We’re not just dumping raw data and saying, “have a good time, AI.” We’re crafting the data structure with intentionality, drawing on our deep knowledge of schools’ data and what AIs want. The user types one sentence, and it can build a full dashboard. That’s amplification: the user’s expertise about their school, combined with computational power they never had before.
5. Earning trust with AI requires transparency.
Because VC Studio generates code, and AI is good at reading code, you can always ask it: what did you do? What formula did you use? You can understand why it returned the results because it’s not a black box. You can find out the chain of transparency. You can ask for clarity, then go back and say, “I don’t like how you did that; do it this way instead.” The mystery doesn’t let you iterate. Transparency does.
6. AI models lack the context intelligence required to be useful.
A generic tool might be able to pull a number, but VC Studio can pull the right number, in context, because we added the context it needs to understand how your school thinks about that number.
We have 25 years of accumulated knowledge about how independent schools structure and name their data, their workflows, their relationships, their reporting needs. That knowledge is what we layer on top of established AI models so that our tools can interpret plain-language requests and return results that are actually right for schools. We’re not training a model; we use off-the-shelf LLMs, and schools’ data is never saved or incorporated into future model versions. We have an enterprise agreement, the same as with all our other vendors. That’s table stakes in our space.
7. Getting it right requires alignment with the model’s strengths and weaknesses.
LLMs are technical tools with strengths and weaknesses. We build in alignment with those attributes, not against the grain of the model. The biggest example: we let the AI write code. Other reporting solutions slapped AI on top of pre-existing report builders as an automation layer; the AI just pushes the same buttons a human could have pushed, faster. But there’s a ceiling. In VC Studio, the AI writes database queries in real time. That unlocks capabilities you otherwise couldn’t access, and it makes results repeatable: once the code is written, I can save it and rerun it without re-involving AI. Repeatable for the user, cost-efficient for us.
Here’s another way to think about it. You have more data inside your SIS than you’ve ever had: enrollment trends, fundraising performance, academic outcomes, student engagement. It’s all there. But getting useful answers can be time-consuming and complex. In the worst cases, people stop asking questions altogether because the effort isn’t worth the time. It’s too easy to become data-rich and insight-poor.
This is the problem we set out to solve with AI. And VC Studio is the first step, not the last. Reporting and visualizations are the starting point, but the underlying platform is designed to grow. Every capability we add will follow the same principles that got us here. We’re just getting started on building the best AI-powered tools to serve the needs of independent schools.