Automation and software
The same data typed into three programs. Approvals chased over email, reports built from separate files, every month. We start from the process eating your time and see what's worth automating, integrating, or building.
Where we start
None of this is a technology problem. It's a process stuck in someone's head. When they're on vacation, the process stops. That's what we fix.
How we work
Not at what program someone wants to buy. What comes in, what goes out, where it gets stuck, who decides.
Often it's an automation between two applications you have. Sometimes an AI assistant. Rarely a new application. That order is deliberate.
Not the manager. The accountant, the receptionist, the technician.
Hours a week, steps by hand, errors. If it hasn't changed, we're not done.
What we build
Invoices go into accounting on their own. Online store orders reach the warehouse without an email. A client's form opens a task for someone. We use your applications' APIs, and code when needed. Every automation has a person alerted if something fails.
One that answers clients' common questions on WhatsApp and opens a ticket when stuck. One that reads order emails into a table. One that drafts a reply for a person to check. Not "AI in the company"; one assistant, one job, one person accountable.
When nothing suitable exists, or what exists costs more than it's worth, we build you an application. Only if we can also maintain it afterward.
Proof
Before proposing an AI assistant to anyone, we built ours. Cipri sits on BitUp's support WhatsApp number. He recognizes the company by phone number, asks the right questions, suggests steps, takes screenshots. When he can't solve it, he opens a ticket in BitUp Helpdesk and hands it to a person. Our clients use him daily.
This isn't to sell AI, but so you know we've run what we propose in production first. How we built it and what we learned
Custom software
The biggest thing we built is ProQ, a platform for automotive suppliers who must prove to clients their process is under control: planning, risk analysis, control plan, part approval file. 29 modules, hosted in our private cloud, in the EU.
Even outside automotive, ProQ is proof we can take software from an idea in a factory to a product maintained for years. We work the same way on a 5-user internal application. Read the ProQ case
What we usually build: internal applications (records, approvals, scheduling) that replace Excel files emailed around; client or supplier portals; integrations between your systems. Web-based, PostgreSQL database, hosted by us or on your side, backup and updates included in maintenance.
What we don't build: native mobile apps, brochure websites, online stores. There are people better at those; we'll say who.
Honest
FAQ
From a few hours, for a link between two applications with an API, to weeks, for a complicated flow with many exceptions. You'll know after we see the process, not before. The first conversation is free and needs no specification.
Usually not. The best automations connect what you already have. We change something only when the old program is the real obstacle.
Depends on what we build; we'll tell you exactly. For AI assistants we use commercial models (like ChatGPT) via API, with minimal data, or models on our side when data is sensitive. We decide together beforehand, in writing.
We do, under a maintenance contract, with the same helpdesk as the rest of your IT. The code is yours, with documentation and access from the start.
Sometimes. We look at the code and tell you straight if it's worth maintaining or rewriting. We don't take over what we can't understand.
Tell us what repeats and where things get lost. In 30 minutes we'll see if it's worth automating, and how.
