How an AI agent in Slack helps our Mobile, Web and Backend teams answer cross-team questions, triage incidents during ticket sales, and spend less time in "sync" meetings.
4 minutes read
20 Nov, 2026
by Favour Emele-Onu

Picture this. It's Sunday evening and you're trying to wrap up a feature before the new work week starts (don't ask me why I didn't just do it on Friday ๐). You reach out to the Backend team, but there's no response. You know this is a blocker, and if you don't get it done, you'll have some explaining to do on Monday morning. I mean, nobody is going to beat you, plus we have a pretty cool boss. Regardless, you really want this done, and everyone else is enjoying their work-life balance ๐.
So, at Jetron, engineering is organised into three pods: Mobile, Web and Backend. Each pod knows its own domain deeply. The problem shows up when you need something from a different domain. Now, you could be a genius and master everything across all the domains, but even then, that's a lot to hold on to. You could also download every repository to your PC so you can read or reference them whenever you need something, but that's pretty stressful on its own.
This post covers the problem that pushed us to bring an AI agent into our workflow, how our engineers use Tosin day to day, and where we've drawn the line on what it should and shouldn't decide.
Before Tosin, questions that crossed domains had one reliable answer: everyone involved had to join a huddle to get in sync. Oh my God, you do not want to know how annoying that can be sometimes, because you never want to be a blocker for someone else. I mean, GTA VI came out recently, and I would lose my mind if someone disturbed me with a huddle or a notification because they needed some information from me. Here are some of the situations where you might have to disturb someone ๐ซฉ
Building on someone else's API. When the Mobile team picked up a new feature, they needed to know which queries or mutations to use and what every property in the response meant. That usually turned into a call with the Backend team that could end up being a really loooooooooooooong one.
Keeping experiences consistent. When the Web team shipped a feature one way, the Mobile team had to implement it the same way so users got the same experience on every platform. That meant another sync, or trying to reach the right people.

Problems that span every domain. At times, we run into issues that can only be solved by someone who understands Mobile, Web and Backend at once. That person has to be available whenever those issues come up, which doesn't scale very well. We cherish everyone's work-life balance ๐งก
The knowledge we needed already existed. It was just spread across pods, codebases and people's heads, and getting to it cost us meetings and patience.
The gap matters most when we're running a sale, especially for an artist with a huge fanbase. During a high-traffic ticket release, knowing that an error happened is never enough. We need to know:
When it happened
Which system reported it
What the evidence actually tells us
Answering those questions quickly requires context from every pod at the same time, which is exactly what our structure made hard.
Tosin is an AI agent that lives in Slack, where everyone already works. Our goal was simple: when a question crosses team boundaries, engineers should have somewhere to start that isn't a huddle or a mention.
In practice, that looks different for each pod:
Mobile engineers ask which query supports a feature, what its response fields mean, or how an existing flow handles an edge case.
Web engineers check whether an unexpected response is intended Backend behaviour or something that needs investigating.
Instead of waiting for the right person to be free, an engineer gets a grounded starting point in minutes, and can bring a much sharper question to a teammate if they still need one.
Tosin has also become part of how we decide which errors deserve attention first. I remember that during one presale investigation, we asked Tosin to group the reported errors into categories and flag the ones that needed further investigation. Just as importantly, its report separated two things that are easy to blur:
That distinction matters. You don't want to spend your energy on things that aren't a priority. Sale periods are usually noisy, and not every error logged during one was caused by it. Separating correlation from cause meant our engineers could focus on the issues that actually affected the presale, rather than chasing every alert that happened to fire at the same time.
Adding AI to our workflow doesn't make every answer correct, and we've designed our process around that.
Decisions still belong to engineers.
Tosin's job is to make sure those decisions are made with better context, not to make them on our behalf.
We brought Tosin on for two reasons: so our team spends less time reconstructing information that already exists, and so decisions across our engineering pods are made with the full picture in view.
The pods still own their domains. Tosin just makes it a lot easier to see across them.
PS: Our engineering team is pretty fast at replying, by the way (maybe even faster than your girlfriend ๐), but Tosin really makes work a hell of a lot easier!
Join Our Mailing List

Discover, share and promote events easily.
Join Our Mailing List
Addresses
Lagos
15 Kososko Street, Off Innocent Street, Suberu, Lagos
Delaware
4835, 1007 N Orange St. 4th Floor , Wilmington, DE, New Castle, US, 19801
Content