busywork
Home / Case studies
Case studies

Our first client was us.

Busywork was born inside our own company: e-commerce and marketplace services with teams in several countries, dozens of clients, accounts on many markets, and a pile of invoices, payments and reports growing faster than the team. These are the processes we automated, how we did it and what changed.

0 APIsAutomated bookkeeping on a cloud accounting system with no API
The problem
The accounting system is slow, cloud-based and has no API. Every month the finance team matched transactions from four banks and two cards by hand, hunted for invoices across several inboxes and posted everything line by line. Month-end closed late.
What we built
A robot on a dedicated server that uses the accounting system through its interface, just like a person. It works when the finance team isn't there: in the evening after close and early in the morning. On top, an AI agent matches payments and documents and proposes the posting with VAT and accounts.
The rule
Nothing gets posted without approval. A person opens the queue, checks, clicks "post" or corrects. Corrections become rules for next time.
What changed
Every morning the report of the overnight run lands on Slack. Finance starts from a prepared queue instead of a blank sheet.
5subscriptions cancelled with the internal platform
The problem
Tasks in one tool, CRM in another, attendance in a third, plus call booking and time tracking. All charged per user, none connected. To know how a client was doing you had to open four of them.
What we built
Busywork OS: work boards with priorities and Slack notifications, a CRM with emails linked to contacts, clock-ins, leave and time off, shared inboxes, call booking and an AI for the team.
How we rolled it out
First an exact copy of what the team already used, with the data imported. Then the changes. The switch happened over a weekend, with the old tools still running just in case.
What changed
We cancelled our CRM, task manager, attendance app, call booking and time tracker. The task tool alone cost about $275 a month. Above all: the data lives together.
Mon 8:00the recap with the actions to take for every client
The problem
Each account lead looks after several clients, each with dozens or hundreds of products. The data was all there, but studying it properly every week for every client was impossible. Drops were spotted late.
What we built
An AI analyst that every Monday reads the week's sales data, compares it with the week before and writes to the account lead: trend, products growing, products dropping, price anomalies and actions in order of priority.
A detail
Weeks run Thursday to Thursday, because marketplace data arrives a few days late. Choices like this, which come from real work, are what make a report useful.
What changed
On Monday morning the account lead knows where to start. The AI has also flagged unusual price increases on its own that nobody had noticed.
0 forwardsinvoices by email reach the queue on their own
The problem
Invoices for software, foreign services and marketplaces arrived in different inboxes, some of them personal. They had to be forwarded, downloaded and uploaded by hand. Some got lost.
What we built
A single mailbox with a dedicated alias, read by an AI agent. Every invoice is recognised, filed and queued for posting. For foreign invoices, the self-billing document is prepared.
The extra check
If an invoice is addressed to a different company it's still imported, but flagged: it's often a supplier error, so it gets fixed right away.
What changed
Nobody forwards invoices any more. Whoever posts them finds them already in place.
1 botfor the team's issue reports, on Slack
The problem
A new internal system brings questions and issue reports every day, from people in different countries. Answering them all would have blocked the people building it.
What we built
A Slack bot the team writes to. If it's a question, it answers. If it's a problem, it prepares the fix and sends it for approval with two buttons. Nothing goes live without an OK.
What changed
Reports have a history, answers come straight away and engineering time goes on new things.

Want to know which of these processes fits your company?

Free assessment