Inside Rival: How We Use What We Build
This blog is part of Inside Rival, a series where our team shares how we use Rival in our everyday work. From favorite tools to AI agents and automations, we're showing you what we're building, how we're using it and what we've learned along the way.
A data expert's view of handing reporting building to a Rival agent: what got faster, what got better, and what to do with a lighter workload.
I've been “The Numbers Guy” on one team or another for six years, and I can tell you there is no such thing as a quick question about data. It usually starts like this: "Hey, quick question. Can you put together a fast report on ____?"
Oh, I wish it was fast.
Find your data sources.
Set the date range.
Find the metrics you need.
Export all of it.
Get all your screenshots.
Put it onto a template, if you're lucky to have one. Otherwise, start building the report.
Compile everything into something visually pleasing.
Write the insights. Make sure they're accurate and comprehensive.
Find a mistake? Backtrack and repeat until it's perfect.
Somewhere around step six, every data analyst has the same daydream. Not a beach. Just a week where the reports make themselves and the hours go to the work piling up behind them: the analysis nobody has asked for yet, the data filter fix you keep pushing to Friday, the question you actually want to answer. Then you move on, painfully, to step seven.
A few hours later: it's finally ready for them, but they needed it thirty minutes ago. And you still have all your other work.
Now multiply that by every report the company needs. The monthly performance report. The weekly version. Daily cadences during new product or campaign launches. The numbers needed before a leadership meeting. The follow-up question to the follow-up question. You might complete your next report in time. But there's always another.
Three datasets, three different answers
The part that made it worse than tedious was that our data sources didn't agree with each other. Our website analytics, our database and our internal dashboard each showed slightly different numbers, thanks to internal filters that had been set by hand, differently, in each one. Check three sources for one answer and you'd get three different responses, all of them defensible. Leadership did the sensible thing and stopped trusting any of them. Which, when you're the one in charge of said data sources, is a great feeling.
So I built an agent on Rival to be the one place the numbers come from. And suddenly things changed.
"Run the September report"
That's the whole ask. Type it into the agent and the report comes back almost before you've let go of the enter key. Same sections we've always used. Every number pulled fresh from our database and Google Analytics. Every up and down from last month already worked out. Internal account activity already stripped out. No screenshots for visuals. No retyping. No hunting for the typo you made while retyping. When I checked the reports we'd built by hand against what the agent produced, they matched. And suddenly I dared to hope that maybe, finally, I'd start getting some time back in my day.
So I kept going, in the same chat, from the same data:
"Can I get this weekly?"
"Which tools got used the most?"
"Where did the traffic jump come from?"
More questions poured out of me. More correct answers came back. This was the thing I'd always daydreamed about. I started making tentative plans for what I was sure would be my future free hours.
The moment of truth
I stared, elated, at a crisp, clean report that used to cost a data analyst two hours they were never getting back. Every section, done. Then I read closer. A dollar figure printed to six decimal places. "September" with no year on it, which the agent took to mean September of last year instead of 2026. And just like that, my plans were dashed. It was just like the other AI tools. Back to normal troubleshooting. Might as well write up a ticket for an engineer and check back in a week.
Or, I could try to fix it myself. That's when it sank in how powerful Rival actually is. I opened the agent's settings, added a new instruction section, and ran the report again. Most of the problems were already gone. For the rest, I went one layer down to the function the agent calls. One small catch: I don't write code. No problem. The function builder's AI was happy to take the wheel. Edited, published, ran it once more. Perfect. The dream was alive. Every table is correct, decimals and all. Every month is the right month.
One data source
Somewhat miraculously, the agent had solved the time crunch of the quick-turnaround report. But it wasn't done. What was sitting in front of me was a source of truth. One filter, standardized, with the list of internal accounts living in a table anyone on the team can edit. Everyone on my team could talk to it. Everyone who used to ask me questions could ask the agent. And when someone brings a different figure from a different tool, the agent says so: both count the same thing, they leave out different accounts, here are the accounts in mine. The argument about whose number is right turns into a thirty-second look at a list.
In one fell swoop I'd reclaimed time, authority and consistency. My free time plans were no longer tentative. I finally had what I needed to get to all the work I'd never been able to get to. And when someone shows up with a different number from a different tool, the agent explains the gap instead of arguing about it. Same metric, different internal accounts filtered out, and here's the list it used. What used to be a debate about whose number was right is now a thirty-second look at a list.
So, what happened next?
The “quick questions” didn't stop. But answering them finally got quick. There's no bottleneck anymore: the agent fields the question at any hour, from the same data set, and gives the team the answer it needs. And with that time, I'm finally getting to the nice-to-haves. The small improvements that help the team in the long run. The projects that make us more efficient. Most of all, it's giving me time to build more Rival agents: SEO agents, data anonymization agents, whatever the week calls for. Every time I hit a problem now, there's a way to solve it with a Rival agent I can build myself.
If you're “The Numbers Guy” on your team
Check out Rival and build an agent. Describe the report you keep rebuilding to RivalBot in plain words, and it will set up your first version for you. Then get ready for a strange week where every request you get flies by with ease.
About the author
Ryan Schmitt is Manager of Insights & Data Strategy at Rival, where he owns reporting and data sources across Rival's client work. While he can't use his newfound free time to blow the dust off his Xbox or watch football highlights, building Rival agents is his favorite new hobby.