There are many agent harnesses, but this one is mine.

Notes from a non-engineer

I'm not an engineer. At a company building agent harness software for programmers typing into black terminal windows, that used to feel like something I should whisper. Everyone around me was shipping code. I was writing docs, planning, talking to people. Useful, but I wasn't really using the thing we make.

So for a while I did what a lot of people quietly do with AI: I nodded along. I'd heard “agent”, “loop”, “harness” and “extension” enough times to nod convincingly in meetings. I secretly googled what a TUI was while everyone else assumed it was obvious. By the time I figured it out, they’d already moved on to APIs, and I was back at square one. I was, to borrow a line from our own What is a Harness? post, someone who was "too embarrassed to ask."

Then I stopped nodding along and did something even better: I opened Pi and started asking.

The harness doesn't care that you can't code. You open it, you type what you want in plain language, and it gets to work. I expected there to be a point where I’d get stuck because I didn’t know enough. Instead, I just kept typing what I wanted. The first time I asked it to read a long, ugly PDF and tell me the three things I actually needed to know, it just did it. I felt a small, slightly ridiculous rush of power.

Under the intimidating surface, a harness is just a tool that turns "I wish someone would do this for me" into "done."

Making it mine
My favorite thing Pi does for me has nothing to do with software. Every morning, it reads through my inbox and my meeting notes, finds the things I actually promised people I'd do and builds me a clean to-do list: one pile for what I owe others, one for what I'm waiting on. I didn't hire anyone or learn to code. I described the job once, and now it's a small routine that runs on my machine, shaped exactly to the way my brain works.

On the days when curiosity strikes and I want to peek under the hood, Pi turns out to be a remarkably patient teacher. I once tasked it with guiding me through crafting a Google Apps Script. It walked me through the whole thing step by step, the way you'd ask your most generous SWE colleague to explain something: in plain language, without making you feel stupid. Somewhere in the middle of one of those explanations, code started to feel less intimidating. Now I build things of my own. Nothing groundbreaking, mostly small tools that make my life genuinely easier, at work and outside of it. From crafting personalized digital invites to managing email engagement metrics and streamlining data with spreadsheet automations.

That's when the tagline finally clicked for me. There are many agent harnesses, but this one is yours. I used to read that as a line for engineers, people who'd tear open the hood and rewire things. Now I read "yours" differently. It doesn't mean you built it. It means you can bend it to your life, whoever you are.

I bent mine around a messy inbox. Someone else will bend theirs around lesson plans, research, or small business bookkeeping.

Why this matters more than my to-do list
Pi is free, open source, and runs on my own laptop. That used to sound like a technical footnote to me. Now I get why it matters: I can shape it around the way I work. What felt genuinely different was that Pi could look at its own code, explain how it worked and help me change it. Almost like the tool could teach me how to rebuild the tool itself. Pi users have shared more than 5,000 extensions with each other, freely giving to others the workflows they've built for themselves..

Much of the AI conversation centers around what the technology will do to us. Sitting where I sit, the interesting part is that I no longer have to wait for someone technical to build the thing I need.

I'm still not an engineer. But I've stopped whispering it. Turns out you don't need to be one to own your tools, you just have to be curious enough to ask and brave enough to keep going.

Jakob, Founders Associate at Earendil