Everything I do, and the one system underneath it all

TL;DR: I do a lot of different things: regulated biotech and pharma writing for clients through Science Inbound, marketing I keep meaning to do to ensure a steady client base, a hobby blog with its own TikTok channel, this brand with its blog and channels and newsletter, and a household with its own paperwork. One agentic AI system runs underneath all of it, for one reason: every one of those areas needs my expertise, and every one of them generates time sinks. The system takes the time sinks. This is the map – how it got this big, why it compounds, and what it’s honestly worth.

The map, as of today

If you laid my working life out as folders, there would be five.

• Client work. The regulated biotech and pharma writing I’ve done since 2011 – freelance through Science Inbound since 2016. The work with the least room for error, the part that helps pay the bills, and the reason everything else has to earn its time.
• Marketing that client work. The Science Inbound site, the content that should be bringing the next clients in. It’s on the list. I don’t have time.
• A hobby blog, with its own TikTok channel. Where a lot of this started.
• This brand: The Julie Kaiser. The blog you’re reading, a TikTok channel, a YouTube channel, and a newsletter.
• The household. Paperwork, projects, a huge vegetable garden, the running maintenance of a family’s systems – which lands on my desk too.

Five areas, one person. For a long time, the only thing they shared was me.

Why one system runs all of it

Every one of those folders has the same shape. My expertise goes in: the science and the writing where applicable, and the judgment about what’s true and what’s good enough to ship. And around that expertise, every area grows its own busywork – formatting, filing, uploading, tracing. Work that takes real time without needing much of me.

That’s the pattern: all of these sections have my expertise going in, but the time sinks are what actually fill the days. So the system’s one job – one system, not five tools – is to take the time sinks. The judgment stays on my desk; that part I wouldn’t hand over anyway.

How the map got this big

Rewind a couple of years and the map was mostly one line: client work, plus the marketing I was already not getting to. Then the hobby blog happened, then its TikTok channel. I loved it – it grew fast, people were interested – and I was spending hours and hours editing two-minute videos. It wasn’t sustainable. I wanted to keep doing this, so I had to find a way to automate the work around it. That’s what prompted everything that eventually became this. I’m still figuring it out.

What I wanted was concrete: speed up the writing, automate the video production. That was the goal. Speeding up the writing took a really long time – I had to learn how to do it well, and I am still learning. Automating video production took a few days. It’s not perfect, but it works about as well as outsourcing it would: I’d still have to do a review round and tell someone which corrections to make. With my current setup I do that review round in my own system, at no extra cost on top of the subscription I already pay. (And as all the other pieces get perfected, this piece will also improve.)

From there, the other pieces kind of fell into place. Now I know how to automate a lot of different things, and they’re constantly popping into my mind: this should be part of the system, that should be part of the system, this is how I can make the system better. That’s what floats through my brain when I’m not here working on it – and every time one of those ideas shows up, I use my handy-dandy voice-memo system to capture it and freeze it, so I know I’ll get to it later.

How I got to AI at all is a different story – fear, mostly, and years of treading water first; the dated, honest version is its own post. The wiki at the center of the setup has its own build story, wrong turns and all. Because using AI and having a system are different things, and I’d been doing the first for a long time before the second existed.

The flywheel I didn’t plan

The processes I build for one area keep feeding and informing the others. The dashboard that ran one client afternoon is the same dashboard running this blog’s pipeline. And the dashboard came the way everything else here comes: I didn’t set out to build one. (Well – I did ask Claude for one, and it wouldn’t build it for me: it said I would just get frustrated with it and stop using it.) It wasn’t until I was shuffling many parallel sessions, and saw exactly the kind of dashboard I needed and exactly what it had to track, that the dashboard came to be. Thinking about what you’d need in the abstract is different from being in the middle of it, realizing exactly what you need when you need it.

A fix I made for one folder keeps paying off in the others. The system post explains the mechanics of why that compounds; what it feels like from inside is simpler:

Some days it makes me think I’m invincible and can do anything.

The confidence is real, but so is the wobble – and both stay, because the confidence gets things built and the wobble keeps me checking the output.

What it’s worth

I’m not going to lie: none of this is perfect quite yet, and it probably never will be. One of those five folders is still mostly a to-do list – but it’s a to-do list with a plan and a strategy, captured in the wiki. The amount of things I’m doing at the moment, not perfect but much better than I could do them alone, makes it worth it in my opinion. At least that’s my calculation.

If you want to see what one day looks like inside the system, the client afternoon is written up. And if the map above looks like your life – expertise everywhere, time nowhere – building a system beside your work is the thing I now do with people.