Walk into the back office of any small business that's outgrown its starting setup, and you'll see the same thing.
The operations manager has fourteen browser tabs open. Shopify admin in three of them. A shared inbox in two more. The accounting tool. A wholesale spreadsheet. A supplier portal. A Slack workspace pinned to the side. A dashboard somewhere. Maybe Trello, maybe a project tool, maybe both.
The owner walks over and says, "we just need a smarter Shopify."
That's almost never what they need.
The thing that's actually broken
The problem isn't any single tab. The tabs all work. Shopify is fine. The inbox is fine. The wholesale sheet is fine.
The problem is that the business lives across all fourteen of them, and the only system holding the business together is the operations manager's working memory. Every supplier confirmation, every photo-verification gate, every wholesale approval, every customer status check, every needs-photo-no-ship reminder — all of it routes through one person remembering to check the right tab at the right moment.
When that person is on vacation, the business gets quietly worse.
When that person is hired away, the business gets noticeably worse.
When that person is just having a Tuesday, things slip.
Making any one tab smarter does nothing to fix this. A faster Shopify admin doesn't help. A nicer inbox doesn't help. A custom dashboard inside Shopify doesn't help. The problem isn't bandwidth in any one surface — it's that no surface knows about all the others.
The shape of the answer
There are two things that actually fix this, and they belong together.
A portal that lives above the surfaces. A single internal place where the operations manager (and the rest of the team, in the right roles) can see what's happening across every relevant system. Inventory state from Shopify. Supplier confirmations from email. Photo-verification status from the warehouse. Approval queues. Order pipelines. The portal doesn't replace any of those tools — it pulls from them, presents them in one place, and lets a human take action without alt-tabbing fourteen times.
Single-purpose bots that handle the routing. One bot per job. Inventory checks. Photo-verification reminders. Supplier-confirmation parsing. Wholesale approval flows. Each one is small, deployable on its own, and replaceable when the business changes. None of them try to be clever. None of them know about each other. They just make the surfaces talk to the right humans at the right moment.
Together, the portal and the bots make the operations manager's working memory the system, instead of a single point of failure.
Why we don't build one big platform
Every couple of months, someone asks us why we don't just build one big "automation platform" instead of a portal-plus-fifteen-bots. Wouldn't it be tidier?
It would be tidier. It would also be wrong.
Single-purpose bots can be shipped in waves. Big platforms ship at the end. If we tried to ship the whole platform, the team would wait until the end of a long build to see any improvement. With single-purpose bots, the team gets the photo-check bot first, then the supplier-confirmation bot, then the wholesale approval bot — each one a complete, owned, deployable improvement to the business. The back office gets better continuously instead of in one big-bang event.
Single-purpose bots fail safely. If the inventory bot has a bad day, the photo-check bot keeps running. A monolithic platform with a bug takes the whole thing down.
Single-purpose bots can be deleted. A bot built for a workflow that doesn't exist anymore — just turn it off. A monolithic platform with a workflow nobody uses is a maintenance burden you can't easily walk away from.
And the portal stays small. It's a window into the systems that already run the business, not a parallel system the team has to maintain.
What this looks like in practice
We've built this exact pattern for businesses where the back office had grown into chaos: a B2B distributor, a small e-commerce operation, an agency-style services firm. Same shape every time. A portal — usually Next.js + Supabase, role-based access, RLS for staff vs. supplier vs. admin. A bench of fifteen-or-so Slack bots, each with one job. A scoped, milestone-based engagement to ship the first wave; subsequent bots scoped individually as the team identifies the next bottleneck.
The result is that the operations manager goes from routing-everything to managing. The bots route. The portal shows the state. The manager handles the parts of the job that actually need a human — supplier relationships, exception handling, judgment calls.
If your back office is fourteen tabs deep and the answer keeps coming up "we just need a smarter Shopify" — it isn't a Shopify problem. It's an architecture problem. The good news is the architecture is small, scoped, and shippable in waves.