Apollo: How Sandbox VR Solved Its Training Crisis

Picture a brand new store opening, and the newest person on the floor has to keep six players moving through a full-body VR game without a hitch. Gear gets knocked loose. A headset drops out of calibration mid-session. And the manager? Nowhere in sight. I came across a case study from the original poster that shows how one company solved exactly this, and I couldn’t stop thinking about how clever the fix is.

The story comes from Sandbox VR, shared by the author along with Danielle Beram, their Sr. Global L&D Manager. If you’ve never heard of them, they run 90+ locations across 13 countries and serve more than 150,000 guests every month. They’re opening a new store every few weeks. That kind of speed creates a very specific problem.

The problem: knowledge that changes month to month

These aren’t small shops. We’re talking 4,000+ sq ft venues where groups suit up in headsets, haptic vests, and body trackers for full-body VR. As the expert explains, all that hardware has to be calibrated and working before guests walk in.

Here’s the tricky part. The operating systems, the tech, and the experiences keep getting updated. So what staff need to know shifts constantly. Now add the games themselves:

• Six people ducking, swinging, and moving for 30 minutes straight
• Gear that gets knocked loose or out of calibration
• Every new store needing staff who can run the floor on day one

Most companies attack this the obvious way. Hire more managers. Train more employees. Repeat forever. It’s expensive, and it never really catches up.

The solution: an AI agent named Apollo

Instead of scaling headcount, Sandbox VR built something different. The creator describes Apollo, a Chatbase AI agent trained on their own internal docs and SOPs. It lives right inside Slack, so it’s accessible to the whole company.

The workflow is beautifully simple. When something breaks mid-session, a staff member just describes the issue in plain words. Apollo replies with step-by-step instructions to fix it, pulled straight from Sandbox VR’s own playbooks.

Nobody goes looking for a manager. Even the newest person on the floor can sort out the issue and get the players back in the game.

The result: floors that run themselves

This is the part I found genuinely impressive. The knowledge that used to live in a manager’s head now answers instantly in Slack. New hires stop being a bottleneck. A fresh employee on their first shift gets the same answer a five-year veteran would.

For a company opening stores every few weeks, that changes the math on growth. You’re not racing to hire and train managers fast enough. The expert’s setup lets the system carry the knowledge, and people carry the experience.

Why this works, and how you can borrow it

The smart move here isn’t the fancy tech. It’s what they fed the AI. Apollo is only useful because it was built from real, specific internal documentation, not generic web answers. That’s the lesson worth stealing.

If you want to try something similar, here’s the pattern the author’s example points to:

1. Gather your existing SOPs, troubleshooting guides, and internal docs in one place
2. Train an AI agent on that material so answers reflect how you actually operate
3. Put it where your team already works, like Slack, so nobody has to hunt for it
4. Let frontline staff self-serve answers instead of interrupting a manager

Any business with fast-changing procedures can use this. Retail, hospitality, support teams, field service. If your knowledge shifts month to month and your people need answers in the moment, a doc-trained agent beats another round of hiring.

I love this one because it flips the usual instinct. The problem looked like a staffing problem. The real fix was a knowledge problem. Solve the knowledge, and the staffing pressure eases on its own.

Want the full story, including what the San Antonio team shared? Check out the original LinkedIn post for all the details. 👇