Build a Business Dashboard with AI: 3 Tips from an Owner-Operator

Introduction

In northeast Texas, Jim and Aimee Eitel raise cattle and dairy goats at Eitel Valley Farms. They practice high-intensity rotational grazing, native grass restoration, and track their funding for multiple conservation programs.

The pair used to keep records on all of their projects and enterprises in a combination of Excel and livestock-tracking platforms, but as their herd grew, this got complicated. “I’m pretty good at Excel,” Jim says, “but it was getting to the point where I couldn’t do all this in Excel.”

He decided to build his own operational command center with ChatGPT. It’s an online dashboard he and Aimee can both open from the field. They use it to check on everything, from the number of days a pasture has rested to the progress they’ve made on conservation projects.

Today, we’re sharing three lessons from Jim’s process. You’ll leave with practical tips you can apply to building your own business dashboard.

Learning Outcomes

By the end of this guide, you’ll know how to…

• Define and describe business goals when working with AI

• Take a step-by-step approach to building AI apps and websites

• Turn questions about your business into dashboard views

Before you dive into the tips below, review the notes from our first AI lesson on using the technology safely and connecting your bookkeeping data to AI tools.

Tip #1: Have AI Work Backwards from a Goal

When Jim decided to create his own command center, he approached working with AI the same way he might plan a ranching or conservation project: Define your ideal end state, then work backwards to figure out what’s needed to get there.

In other words, instead of trying to give AI software or design specifications, Jim recommends starting with a goal. What do you want to know about your business, and how do you want to receive that information?

Here’s an example snippet from his conversation in ChatGPT:

I want a mobile-friendly, web command center for my farm. When Aimee or I open it, it should immediately tell us what needs attention, which animals are being watched, where the cattle have been, how long each pasture has rested, what work is coming due, and what has changed.

Jim focused on explaining the outcomes he wanted for the dashboard in plain language, then let ChatGPT translate that into the components of a working website. “I’m going to the end state of the data point I need, and reverse engineering with AI to get that information. I’m also guiding AI on where I want it to pull data from.”

For Your Dashboard

Here are a few questions to help you define the end state for your business dashboard. You can use your answers to put together an initial prompt for ChatGPT.

• Who will use the dashboard, and where? Whether you plan to use the app at your desk or from your phone might change how AI builds it.

• What are the numbers you’re constantly checking on? These could be related to running your operation (like animal health or pasture rest periods) or doing its finances (like cash flow or unpaid invoices).

• What are the tasks you’re constantly working on? Maybe there are milestones for funding programs you need to track, or you’re often reminding yourself to reconcile the books.

• Where does your most important business data live? Gather a list of the accounting software, spreadsheets, operational software, or other places AI may need to pull data from.

Then, you can put this together in plain language. A simple prompt could look something like this:

I want a mobile-friendly dashboard for my [type of operation]. When [user] opens it, it should immediately tell us [answers to the questions above]. Ask me questions about how we operate and where to find this data before you start building.

Tip #2: Create an Imperfect First Version

After Jim laid out the goal for his command center, he paused there. He let ChatGPT build out a basic first version, even though it didn’t include everything he’d eventually want in the dashboard.

He took that first version of the command center and just started using it on the farm. This helped him quickly identify problems and pick out what was still missing from the dashboard. Then, he would note these gaps, go back to his laptop, and give AI one problem to solve at a time.

For example, Jim would write prompts like:

• Aimee needs to be able to log in separately and update the same records.

• I need species, class, sire, and dam fields in the livestock records.

• I need to filter animals by species and sort the tables by any column.

• Pastures need to show the last grazing date and automatically calculate days of rest.

• Add a farm calendar and connect tasks, grazing moves, calving estimates, and reminders to it.

• I need to see the progress on our conservation projects, both from an operational and a funding tracking perspective.

• I need to see monthly and year-to-date rainfall.

• Make sure Aimee and I can see the same information in our browsers rather than saving separate copies on our individual devices.

ChatGPT generated a new iteration of the command center after each of these requests, which Jim would take back to the farm to test again. This cycle of feedback took time and effort; as Jim put it, “I can’t tell you how many times I said: No, I want this.” But over the course of a few months, it produced 18 versions of the dashboard that were progressively more stable and useful for Jim and Aimee.

For Your Dashboard

As you test the first version of an AI-built app, keep a running list of anything that doesn’t work or feels difficult to do. Here are some questions to help you identify what to ask AI to fix:

• Convenience: Did the dashboard actually tell you information you wanted to know in the field, or did you still have to open another app or spreadsheet?

• Incompleteness: Is there a number, task, or deadline you expected to see that isn’t there?

• Accuracy: Spot-check a few specific figures against their original data sources. For example, compare any financial metrics that AI pulled with what you see in Ambrook.

• Reliability: If you entered new data into the command center, did those changes save and update the way you expected them to?

For any issues or missing items that come up, turn them into a request for ChatGPT. You can describe the problem you need solved or the outcome you’d like to see. For example: I need to see unpaid invoices sorted by oldest first.

Tip #3: Ask AI to Reset

In some cases, as AI kept producing new versions of Jim’s dashboard, it felt like it was creating new problems instead of making the artifact better. Jim ran into one issue where, after several rounds of asking ChatGPT to add new features, certain parts of the command center stopped working altogether.

For example, the web app was supposed to allow Jim and Aimee to log in as individual users, enter new data about their farm and animals, then sync this information for all users. But he started having trouble with logins and data that wouldn’t save.

Here’s how Jim fixed this reliability problem:

1. Pause new features: For the time being, he stopped asking ChatGPT to add new functionality to the dashboard.

2. Audit AI’s prior work: He requested that AI review what it had built so far. Inventory everything I’ve asked you to build up until this point, then diagnose why login and data storage aren’t working.

3. Rebuild what isn’t working: He then had AI rebuild certain parts of the command center. Make sure to preserve any existing records or data in the command center so far, then rebuild the sections that are causing login and data storage issues.

4. Test the rebuilt system: He asked AI to go through the full workflow that he and Aimee would typically take in the command center. Test the features in the rebuilt command center to see if you can get through a full workflow: try logging in, entering a new piece of data, editing the data, navigating through the tabs, saving data, and syncing it to all users.

By having AI pause, rebuild, and retest the command center, Jim got to a version that worked reliably and that he could continue adding features to.

For Your Dashboard

The more specific you can be with ChatGPT about what’s wrong, the faster it can diagnose and fix the problem. Whenever you notice something isn’t working in your custom dashboard, take note of:

• Intent: What were you trying to do when the software broke?

• Expected vs. actual behavior: What did you think was going to happen when you took an action in the dashboard, and what did the software actually do?

• Technical specifications: What device or browser were you using when you noticed the problem? Which tab or section of the dashboard were you using?

• Consistency: Have you noticed the broken feature multiple times, or does the problem happen only sometimes?

• Timing: Did you start noticing the problem only after ChatGPT added a certain feature?

This way, you can give ChatGPT a specific description when problems come up: When I [take an action] in [a section of the dashboard], I want [expected result], but instead I see [what actually happened]. This issue started coming up after [recent change]. Investigate the problem.

The Result: AI as a Tool to Fill Business Gaps

For Jim, the point of using AI to build custom software is to meet specific needs for his business: an easy way to track the health of both his cattle and goats, or a single place for him and Aimee to collaborate on their operation. “It may not replace another tool we used, but it filled a need that we had.”

Being able to pull different parts of his operation into one place (from managing livestock, to running conservation projects, to tracking pasture rotations) has been the most useful part of the AI-built command center. “Having all this integrated, it’s pretty cool… It’s game-changing.”

Resources to Dive Deeper

• Learn how to use AI for more specific business tasks, like reviewing your ledger.

• Hear another Ambrook team member and full-time rancher, Haley Potter, talk about how she uses AI for ranching on the Ranching Returns podcast.

FAQs

Do I need to know how to code to build a dashboard with AI?

You don’t need to know how to code; Jim built his command center by describing outcomes in plain language and letting ChatGPT work on the technical build. The important part is having a clear business goal, so you can evaluate whether AI’s generated result actually works well for your operation.

What should a farm or small business dashboard track?

You can use a business dashboard to track data that would usually be spread across different places or software platforms. For example, if you want to check on your cash flow and AR/AP every morning, but also keep up with conservation project deadlines in the same place, you could bring that financial and operational information together in a custom dashboard.

Can I connect my books to ChatGPT?

Yes. You can connect accounting software like Ambrook to ChatGPT through an MCP connection. We wrote more about how this works and how to set it up in our lesson on building a financial health report with AI.

Will AI change data in my dashboard automatically?

AI may be able to update data in a custom dashboard automatically if you connect it to relevant data sources (like your accounting software, or operational platforms you use) and give it permission to edit the dashboard. If you don’t want AI updating your custom dashboard, instruct ChatGPT not to make any changes without your approval. We wrote more about keeping your data safe and controlling AI’s editing permissions in our first lesson on AI for financial management.

Can more than one person use the same dashboard?

It depends how ChatGPT built your custom website or web app. For example, Jim’s first version of the dashboard saved data only on his own laptop, so updates didn’t sync between his device and Aimee’s phone. If you want the dashboard to sync across multiple users, try requesting that functionality in your AI chat. Here’s how Jim prompted ChatGPT: Make sure Aimee and I can see the same information in our browsers rather than saving separate copies on our individual devices.

This resource is provided for general informational purposes only. It does not constitute professional tax, legal, or accounting advice. Consult a qualified professional about your specific situation.