Why Spillway Exists

Over the last few decades, email has quietly become an everything box.

It is where the schedule for your kid's soccer practice arrives. It is where the collaborative document you need to edit lives. It is where you will find travel confirmations, receipts, grant deadlines, student questions, appointment reminders, bills, newsletters, and hundreds of other things that somehow all end up in one place.

The problem is not simply that everything comes to one place. The challenge of everything coming to one place is more than just logistical—it is also an attention disaster.

However much we may think otherwise, humans are terrible multitaskers. Every time we switch between projects, we spend time and mental energy recalibrating ourselves to what is coming next. Those costs add up surprisingly quickly.

Email is basically perfect kryptonite in this environment.

When you open your inbox, you do not just see one thing. You see the email you intended to answer, the newsletter you forgot to read, the receipt you should probably save, the meeting you need to schedule, the message from your child's school, and the research document someone just shared. Before you have replied to a single message, you have already changed mental gears half a dozen times.

Even looking for one email is often enough to distract you with three others.

So instead of spending your energy actually doing the task in front of you, you spend it deciding which task you ought to be doing. By the time you have made that decision, you have often lost the focus you needed to do it well.

What I wanted was not simply a better email client.

I wanted a way to work that let me think about one thing at a time.

Reply to student requests. Review newsletters. Deal with scheduling. Handle documents. Track commitments.

Each of those requires a different mindset. The inbox should not force me to bounce among them every thirty seconds. It should help me stay focused long enough to actually finish something.

AI looked like the answer—but there was a catch

At first glance, AI looks like the perfect answer.

Modern language models are remarkably good at sorting information, recognizing requests, identifying dates, pulling out commitments, summarizing conversations, and grouping similar messages together. Those are exactly the kinds of tasks that make an inbox overwhelming.

But there is a catch.

Most AI systems do that work by sending your email to someone else's computers.

Sometimes that is perfectly reasonable. Sometimes it is not.

Depending on what you do, your inbox may contain student records, confidential research, legal discussions, corporate information, or personal conversations that simply should not be sent to a cloud service. And because email arrives in an unpredictable order, it is difficult to guarantee that sensitive information never gets swept up in the process.

Right now, that is often the tradeoff.

Either expose your information in exchange for AI assistance, or protect your information and do all of the sorting yourself.

I do not think that should be the only choice.

If AI can do much of this work directly on your own computer, then your email never has to leave your machine simply to be organized.

That is the underlying idea behind Spillway.

Use AI where it helps. Keep your data where it belongs.

Why I built it

As a university professor, the father of two elementary-school-aged kids, and the person who keeps our household running, I found myself dealing with about the worst possible version of this problem.

My inbox was not just work. It was not just home. It was not just newsletters or travel or collaborative documents. It was all of them, constantly interleaved.

For years, my strategy was basically triage. I would deal with the most pressing messages from the dozens that arrived every day and hope the less urgent ones would still be there when I finally had time.

Every few months, I would look at the enormous pile of unanswered messages and, somewhat mournfully, mark hundreds of them as read because I simply did not have the bandwidth anymore.

There were entire days when I intentionally avoided checking email because I knew that opening my inbox was the fastest way to guarantee I would not get any real work done.

After trying a number of commercial email clients, I eventually came to the conclusion that none of them really worked the way I think.

So I decided to build the one I wanted.

Spillway reflects how I actually work

I knew a lot about the psychology of attention. My research and teaching have spent years examining how people process information, make decisions, and navigate increasingly complicated environments. I also knew the kinds of information I needed to pull out of email.

I did not want to lose a date six months in the future because it disappeared into my inbox.

I wanted to collect requests that needed replies instead of hoping I would remember them.

I wanted a place to review newsletters when I chose to review them instead of when they happened to arrive.

I wanted documents to be easy to find again.

I wanted the software to learn how I organize email instead of forcing me to adapt to someone else's system.

Just as importantly, I knew the interface mattered as much as the AI.

The goal was not simply to automate sorting. It was to reduce the amount of attention email demanded so that I could spend more time doing meaningful work and less time deciding what to do next.

One workspace, different boundaries

Another important design decision was that I did not want to separate work and home completely.

That has never reflected how my life actually works.

My children's schedules affect my work calendar. Travel affects both. Some projects involve personal and professional commitments at the same time.

Switching between separate inboxes never really solved the problem because each inbox was still an everything box.

What I wanted was one workspace that respected the real boundaries between different kinds of information.

That is one of the reasons local AI became such an important part of the project. Different accounts deserve different privacy rules, even if I want to see them together.

Building my own lighthouse

There is a familiar collective-action problem in software.

Many people wish a particular tool existed. Many people would benefit if someone built it. But building durable software takes years, so it is often easier to wait for someone else to begin.

Sometimes the only practical answer is to build your own lighthouse.

That is what Spillway is for me.

It began as software I needed for my own life.

Today, it is the email client I use every day.

I am releasing it because I suspect I am not the only person trying to navigate this particular stretch of coastline.

If it turns out to help other people too, then building the lighthouse will have been worth it.

About the creator

Spillway was created by Josh Pasek, Ph.D., Professor of Communication & Media and Political Science, Research Professor at the Center for Political Studies in the Institute for Social Research, and Core Faculty in the Michigan Institute for Data & AI in Society at the University of Michigan.

His academic work examines new media, political attitudes and behavior, public opinion, survey methodology, emerging technologies, and data science. He also develops and maintains research software and teaches about the social consequences of generative AI.

Spillway grows directly out of that combination of interests: building practical software while treating privacy, governance, institutional trust, and the social consequences of AI as design requirements rather than afterthoughts.

Read Josh Pasek's full biography.

Questions, feedback, and beta inquiries may be sent to software@joshpasek.com.