Mar 2, 2026 Mar 2, 2026 2026-03-02

In the last 2 weeks, I shipped not one, not two, not three, but exactly four fresh iOS apps: MosquiGo, AI Kiddo, Zen Tales, and Just 5 Words. Each one a small world of its own — feature flows, design decisions, App Store rejections, support emails, the whole shebang. Oh, and I’m still maintaining addTaskManager on top of that, because apparently I enjoy keeping myself busy.

I also have a one-year-old. He is very opinionated about my keyboard – and, in general, about my so called free time.

Somewhere in between all of this, I built AIGernon. Not as a side project, not as something to show off. I built it because I genuinely needed help. Someone — or something — to handle the boring, repetitive stuff so my brain could stay where relevant things need to happen.

---

The Problem With Running Everything Alone

In general, being a solo founder is fine. The work itself is rewarding. Writing code, designing experiences, thinking through product decisions — that’s the part I signed up for – and thoroughly enjoy. What is not that fine is how much of my actual day gets eaten by everything around the work.

Like re-reading my own text to catch typos at midnight. Trying to remember where I left off on a project three days ago because life got in the way. Context switching between four different apps, a blog, a coaching practice, and a baby who decided 5am is a perfectly reasonable start to the day.

My brain is not the bottleneck. My attention is.

I’d been thinking about this for a while. I use my own ADD framework — Assess, Decide, Do — to structure how I work. The idea is simple: most productivity problems aren’t about laziness or lack of discipline. They’re about cognitive confusion. You don’t know what mode you’re supposed to be in. Are you gathering information? Are you supposed to be making a decision? Should you just be executing and stop overthinking it?

The framework helps. But applying it consistently, across dozens of tasks and projects and half-finished ideas, still takes mental energy. I wanted something that could carry some of that weight.

So I started building AIGernon. Named after Daniel Keyes’ – Flowers for Algernon short story, if you’re wondering. It felt right.

---

What It Actually Does

AIGernon isn’t a chatbot I talk to when I’m bored. It’s more like a work partner that’s always available and never gets tired of my questions.

It runs on the ADD framework natively. Which means every interaction is tagged — Assess, Decide, or Do — and the system understands what kind of response I actually need. If I’m in Assess mode, I need information and options. If I’m in Decide mode, I need clarity and commitment. If I’m in Do mode, I need to stop thinking and start moving. That distinction sounds philosophical until you’re at 11pm staring at a task list and you can’t figure out why nothing is getting done.

I built a custom projects module that keeps context alive across sessions. That one alone saves me probably 10 hours a week of re-explaining where things stand.

There’s a coaching module in there too – like support for my coaching activities: where we are in the practice, what’s pressing, what to discuss in the next session, etc. I have coaching clients for more than 8 years, and there’s a lot of accumulated history. It’s good to have everything in one place.

As you can see, it’s all very customized, it’s literally the shape of my work, poured into an LLM which knows Assess Decide Do.

For a while, all of this only lived in the terminal. And it can still be accessed like this, but, honestly, it feels a little cold. I was fine with that — I built it for me, after all. But there’s something very supporting in having an actual visual overview of your mental map, it shed light on it.

---

Now It Has a Face

The new web interface changes the texture of using it entirely.

Real-time responses, streaming text, typing indicators — the basics. But the part I actually use is the sidebar. It tracks how I’m splitting my day across the three cognitive realms and shows it as a simple color-coded bar. Assess, Decide, Do.

It sounds like a gimmick. It’s not. There have been days where I looked at that bar at 4pm and realized I’d spent the entire day in Assess mode — reading, researching, collecting information — without making a single actual decision. That’s useful data. The kind you don’t normally have access to because nobody’s watching how you think.

The backend is around 80% done. Very usable, full ADD integration, coaching, projects, proper security. If you’ve seen OpenClaw, it’s in that territory — just leaner and a lot more personal. The frontend is at 90%, which in my case means I use it every single day, but I wouldn’t call it production-ready with a straight face.

Everything lives in the same monorepo. Docs are updated. If you’re curious or you want to poke around, it’s all there — AIGernon on GitHub. If you want to evaluate an agent that’s 90% thinner than OpenClaw, but packing a very strong punch, install it, play with it, try to break it and contribute, or fork it and make your own.

I’ll keep building it for as long as it keeps helping. That’s the only metric I have, and honestly, the only one I need.