Request a demo
All posts
Dark blocks with fine outlines, the systems, fitted together like a plan. Bundles of fine lines arrive through the channels between them from the left, the top and the bottom, and leave together to the right, glowing orange to blue.

We've designed the systems. But what about the work between them?

Alex Mee3 min read

For decades, we've been adapting the way we work to fit the software we use. New systems come with new processes. Teams change how they capture information, collaborate and make decisions, often because that's how the technology expects them to operate.

As a product designer, I've spent much of my career trying to make that relationship easier. Understanding people's needs, identifying friction and designing experiences that feel more intuitive. But even the best-designed software has its constraints.

Ultimately, we're still asking people to interact with technology on its terms. AI presents an opportunity to turn that relationship around.

The work happening between the systems

Most organisations don't operate within a single piece of software. Information lives across platforms, departments and teams. People move between systems, reconcile data, chase updates and coordinate decisions.

Over time, they develop ways of making everything work together. A spreadsheet becomes the bridge between two platforms. An email becomes an approval process. A weekly meeting exists because information isn't readily available elsewhere.

These workarounds can become so embedded in day-to-day operations that we stop questioning why they exist. And while each system might be performing exactly as intended, the experience of working across them can be anything but efficient.

Historically, addressing this friction has often meant introducing another tool, implementing a new platform or redesigning an entire process. But, what if we could improve the work without asking everyone to start again?

Four systems, crm, finance, operations and hr. Between them a person holds it all together, chasing each system by hand, with a spreadsheet bridging two of them, approval by email, a weekly status meeting and figures reconciled by hand.

Building for the work between systems

This is where we see an opportunity to do things differently.

At Modolab, we're not building another collection of products and interfaces for people to learn. We're building AI agents that work in the background, connecting the systems organisations already depend on and taking on the work that happens between them.

The same four systems, where they were. The workarounds have gone: an agent between them gathers information, coordinates activities and handles repetitive work, and passes the decision up to a person, who keeps the judgement calls.

We're working with people across organisations to understand where their time is being lost, where processes break down, and where the greatest opportunities for improvement exist. Sometimes AI will be the answer. Sometimes it won't. We're already seeing that in our recent Modosignal engagements, and we're not afraid to say so.

Where AI is the answer, our aim is to build agents around the reality of how those organisations operate. Gathering information across systems, coordinating activities and handling repetitive work, while leaving decisions that require human judgement in the right hands.

The ambition isn't to replace the tools people use. It's to remove the unnecessary work involved in making those tools work together.

Understanding the work before changing it

But as many of my fellow designers will appreciate, there's a challenge in building technology around how people work. First, we need to understand what that work actually looks like. There's often a difference between how a process is documented, how people describe it, and what happens in practice.

A documented process runs straight through five steps, from request to close. Drawn over it, one strand per case, is what happens in practice: some cases are sent back for rework, some skip approval when urgent, some are tracked in a spreadsheet, and one exception leaves the map, circled in orange: never written down.

People develop shortcuts, workarounds and informal ways of getting things done. Processes vary depending on the situation, the person and the exceptions they encounter. Much of this behaviour becomes so familiar that people don't necessarily think to mention it. This is why human-centred research has always been so important to good design.

Interviews, workshops and observation help us understand not just what people do, but why they do it, and the context in which their decisions are made. But, as we design systems capable of taking action independently, we need to understand those behaviours with greater depth and consistency.

Something I'm extremely excited to talk more about soon is that we're developing new ways to observe how work unfolds in practice, capturing the patterns, variations and exceptions that traditional research alone can struggle to uncover at scale. An overlooked exception in a traditional interface might create frustration. In an autonomous process, it could mean the wrong action being taken without anyone immediately noticing. Those spreadsheets, email chains and informal handoffs aren't simply inefficiencies waiting to be automated. They're evidence of how people have adapted to make their organisations work.

Understanding why they exist is what allows us to determine which activities can be safely automated, which processes need redesigning, and where human judgement must remain.

The opportunity isn't to give people another way to interact with technology. It's to understand their work well enough that they no longer have to hold all the pieces together themselves.