PixieBrix Blog

How Our Support Copilot Cuts Chaos in HubSpot, Slack & Gmail

Written by Brittany Joiner | Sep 2, 2025 6:01:08 PM

👋 Hi everyone, PixieBritt here. You might have seen me in our developer community, our YouTube channel, or even our site's live chat!

I wear multiple hats at PixieBrix: mod developer and support lead. The roles are pretty intertwined for technical products like ours!

This means that on any given day, I’m juggling user questions in HubSpot, troubleshooting email threads in Gmail, and community questions in our Community Slack. Not only am I juggling all the channels, but there’s a lot of context switching between the types of questions I get.

Sometimes, it’s a user reporting a bug, sometimes it’s a request for a new feature, but more often than not, it’s a question that leaves me wondering, “Is this a product bug? Or does the user have something configured incorrectly, and I need to sort it out?”

Debugging that answer is no small feat. It requires jumping across a variety of links and tabs to look up user info in MixPanel, DataDog, and our internal tools, then finding any related GitHub issues, help articles in Notion, or even public documentation.

If I can find all that info, sometimes I can solve their problem, but when I can’t get all the context, I usually have to loop in an engineer and give them as much context as I have, and they have to find the rest.

This takes up so much time and leads to context switching left and right for everyone on our team.

There Had To Be a Better Way...

We needed to find a way to make this more efficient, and all our research led us to AI tools for deflecting cases, not solving them, so we built our own. The idea was that we’d use AI to categorize a ticket then open a sidebar that showed the related links to troubleshoot and solve a case, right next to the case.

Once we got it in place, I remember our first ticket coming in via HubSpot, and it was a doozy.

Usually, I’d scan the message, dig through docs, open our internal tools, and try to piece things together. This time, Copilot instantly surfaced the intent behind the request and pulled up relevant documentation with examples.

I was able to respond quickly since I was already starting to understand the problem, which was a 72% improvement compared to our average first response time.

Solving, not deflecting

The user needed help performing a specific action we don’t have in the docs. Good idea to write a doc about, but this was a pretty niche request, as most of ours are, which is why deflection tools won’t work for us. Instead of burning minutes (or hours) gathering context, I had key information ready for me, and clickable action items to start solving this case:

I didn’t have to stare at the case like a deer in the headlights and wonder: "Where next?”

The sidebar was my map.

I could join this user’s team and impersonate them with a click, helping me get under the hood and see exactly what was happening.

I often use ChatGPT to help me with user requests, especially when I need a JavaScript function for transforming output from bricks. So, I used our Chat Prompts, which already had the context of what format the user needed, to have AI suggest a function to perform exactly what was needed.

Our Copilot even solved the blank page issue for me by providing templated responses that I could adjust to respond to the user.

I could unblock the user fast, and most importantly, set the right tone from the start.

And when I was out of the office? A teammate was able to jump in cold. Thanks to Copilot’s automatic summary of the conversation so far and not just the initial question, they were up to speed in minutes and ready to help.

Every step of the way, Copilot nudged me toward the next best action.

The Results

By the time we solved the issue, the customer had not only been unblocked but also thanked us for the thorough response!


Even though this was just a single case, it was a significant win for our whole team:

  • No engineers had to be looped in (a rarity for most tickets)

  • I’m empowered to handle cases, even though no two are ever the same

And it’s not just a vibe of feeling great. We’re seeing measurable changes: faster responses thanks to instant suggestions and higher customer happiness scores after interactions.


Why This Matters

Support work is messy. It happens across tools, across teams, across time zones. But the right Copilot doesn’t replace human judgment; instead, it gives you superpowers to focus on what matters: helping people and solving problems.

Want to try this tool out? We’re inviting folks to apply to our Early Access Program now to receive free white label customizations and hands-on setup to start getting value from a tool like this on your team.

Join Now