We Gave Our Team an AI Voice Assistant — Here's What Changed

Every morning at our dental practice, the same ritual played out. Our practice manager would arrive, sit down, and start clicking through systems. Check the day's schedule. Scan for cancellations. Pull up the lab case tracker. Open emails. Review the roster for tomorrow. Thirty minutes of bouncing between screens before she could actually start managing the practice.

It wasn't inefficient because she was slow — she's brilliant at her job. It was inefficient because the information lived in five different places and she had to go find it every single time.

So we built something to bring the information to her instead. We call it Spark.

What Spark actually is

Spark is a voice assistant that's connected to our practice systems. Not a generic smart speaker. Not Siri with a dental degree. It's a purpose-built AI that knows our practice, our software, and our workflows.

You talk to it like you'd talk to a colleague. And it talks back with actual answers pulled from real data.

Here's what a typical morning sounds like now:

• "What's on today?" — Spark reads the day's appointment list, highlighting anything unusual like double-bookings or long procedures
• "Any cancellations?" — it checks for gaps in the schedule so the team can start filling them early
• "Read me the lab cases due this week" — it pulls from our tracking system and lists what's expected and what's overdue
• "Any urgent emails?" — it summarises flagged items so nothing important gets buried
• "Who's rostered on tomorrow?" — it reads back the roster, including any recent changes

Five questions. Five answers. The whole morning briefing done in about five minutes — while she's making a coffee.

The part where everyone hated it

We'd love to tell you the team embraced Spark from day one. They didn't.

The most common reaction was some variation of "I don't want to talk to a computer." Which is fair. Talking to an AI out loud in a shared workspace feels strange at first. People feel self-conscious. They'd rather just click through the screens the way they've always done.

A couple of staff members tried it once, got a slightly wrong answer (we were still fine-tuning the lab case integration), and wrote it off entirely. "See? It doesn't even work properly."

We didn't push it. We just left it available and kept improving it. The practice manager kept using it because she could see the time savings. A couple of others started using it for quick questions during the day — "When's Mrs Thompson's next appointment?" — because it was faster than opening the software.

About two weeks in, something shifted. The staff who'd been skeptical started noticing that the people using Spark just... knew things. They knew about the cancellation that came in overnight. They knew about the roster change. They weren't caught off guard by the emergency appointment that got squeezed in.

By week three, everyone was using it. Not because we told them to. Because they didn't want to be the one who didn't know what was going on.

Where it shines

Spark is genuinely excellent at three things:

Quick information retrieval. Any question that has a factual answer sitting in one of our systems — Spark can get it faster than you can open the software. Patient's next appointment, today's schedule, whether a lab case has arrived. These are five-second interactions that used to take 30 seconds of clicking.

Hands-free queries during procedures. This was a use case we didn't anticipate. Clinicians started asking Spark questions mid-procedure when their hands were occupied. "What's the next patient's medical history flag?" No need to de-glove, walk to a screen, look it up, re-glove, and come back. Just ask.

Morning briefings. The original use case, and still the best one. Getting a complete picture of the day ahead without touching a keyboard. It's become part of how we start every morning.

Where it falls short

We're not going to pretend Spark does everything. It doesn't, and it shouldn't.

Clinical judgement is off the table. Spark will never tell you whether a treatment plan is appropriate or whether a patient needs a referral. That's not what it's for. It retrieves information — it doesn't interpret it clinically.

Complex multi-step requests trip it up. "Check tomorrow's schedule, find the gaps, cross-reference with the waitlist, and book the best matches" — that's too many steps chained together. You need to break it down or just do it yourself. We're working on improving this, but it's an honest limitation right now.

Sometimes you just need to see it. A voice summary of the schedule is great for a quick overview. But if you're trying to rearrange six appointments to fit in an emergency, you need the visual. Spark gives you the snapshot; the software gives you the control.

Knowing these boundaries matters. We've written before about being realistic with AI — the tools that actually work are the ones built around specific, well-defined tasks. Spark works because we didn't try to make it do everything.

The numbers

We tracked the impact over the first couple of months:

• Morning prep time: 30 minutes down to 5. The practice manager gets the same information in a fraction of the time.
• Fewer surprises. "I didn't know about that" moments dropped noticeably. Staff are more informed because the barrier to checking is so low — just ask.
• Staff actually like using it. This might be the most important metric. A tool nobody uses is a tool that doesn't work, no matter how clever it is. The team voluntarily uses Spark multiple times a day.

Twenty-five minutes saved every morning doesn't sound like much. But that's over two hours a week. Over a hundred hours a year. And that's just one person's morning routine — it doesn't count the ad hoc queries throughout the day from the whole team.

This isn't just a dental thing

We built Spark for our practice, but the problem it solves is everywhere. Any business where staff need to check multiple systems to get a picture of their day is a candidate for this kind of tool.

Think about it:

• Trades — a sparkie or plumber asking "What jobs have I got today?" and getting a rundown from their job management system
• Hospitality — a restaurant manager asking "How many covers tonight?" and "Any dietary requirements flagged?" before service
• Professional services — an accountant asking "What deadlines do I have this week?" and getting a summary from their practice management software
• Retail — a store manager asking "What stock is running low?" instead of running a report

The pattern is the same everywhere: information trapped in software that people have to manually go find. A voice assistant doesn't replace the software — it gives you a faster way to get answers out of it.

What we'd do differently

If we were rolling this out again from scratch, we'd change two things.

First, we'd start with fewer integrations. We connected Spark to everything at once — schedule, emails, lab tracking, roster. It would have been smarter to start with just the schedule, get that rock solid, let people build trust, then add more. Each new integration is a potential point of failure, and early failures kill adoption.

Second, we'd involve the team earlier. We built Spark and then presented it to the staff. If we'd asked them "what do you spend the most time looking up?" before we started, we would have prioritised differently. The features they cared about most weren't always the ones we assumed.

The bottom line

Spark isn't magic. It's a well-connected assistant that saves time on repetitive information retrieval. But that's exactly why it works — it does a specific thing, it does it well, and it fits into how people already work rather than asking them to change.

The real win isn't the time saved. It's that our team walks into each day already knowing what's ahead. No surprises, no scrambling, no "why didn't anyone tell me?" That kind of awareness changes how a practice runs.

If your team spends their mornings clicking through systems trying to get up to speed, there's a better way. Get in touch and we'll show you what a voice assistant could look like for your business.