I'm looking for 30 programmers to test one idea about daily progress notes

For the last 17 months, I've been building GoalWatch on nights and weekends. I'd like to find out whether it actually works, and I can't answer that on my own.

The idea

Most programmers already write a daily standup. In my experience, it's three bullets in Slack, and almost nobody reads them. When I work alone, on a personal project or remotely, I don't write one at all. The work happens, but nobody sees it, including me a week later.

GoalWatch is a daily standup for one person. Each day you write a short update on what you worked on. It can be one sentence or a few paragraphs. A group of AI characters, called Watchers, then react to what you actually wrote. They don't summarize it or score it. Each one has its own name, class, backstory and temperament. Some are on your side, some are indifferent, and one of them wants you to quit.

To best demonstrate how it works, here one of my last updates:

I know what I have to do. I need to make a strong coffee and sit down in total focus and write it down within one hour. That would work better than anything. But currently, I am in a state caused by external factors, I can't drink coffee. Also, I don't have that much time to dedicate an entire hour block today. In the evening, I might be too tired for creative thinking.

I was kind of hoping that easy solution would magically pop into my brain one day. But this doesn't seem to be true.

Watch reactions:

Keldric (BERSERKER):

The frost never leaves my bones. Baby steps forward with external constraints. A soldier's march through hostile terrain. The ledger records your daily resistance, not your velocity. That, at least, carries weight in the accounting of war.

Solara (SORCERER):

I sense your dimensional shift—between that ideal creative state you know works (the strong coffee, the hour of focus) and the reality of constraints that won't bend to your needs today. The Prismatic Engine recognizes this pattern.

Arcanist Evren once waited three weeks for inspiration to strike while her project deadline approached. When she finally sat in her workshop for twenty scattered minutes with tired eyes, the solution came not as lightning, but as embers carefully tended despite exhaustion.

Elara (PALADIN):

Return to the foundation. The roll records your two-day departure and your return alike. External factors are accounted for, not forgiven. The pace has broken. Now it must be rebuilt—one brick, one day.

Evil also speaks no you daily to make quitting more painful:

Syrin (EVIL):

That's a brick, yes, but what if—what if—what if you took all your skills and applied them to that OTHER idea you had in the shower three weeks ago? That one had so much more potential than this dusty old concept!

Why the game parts are there

I know "productivity app with RPG characters" sounds like a gimmick, so here's the reasoning behind each piece.

Characters react so you don't have to judge yourself. When I reread my own notes, I judge them: good day, wasted day. That judgement is where the guilt comes from, and guilt is what makes me stop writing. When a character reacts instead, I can look at my week with some distance. A few characters with opposing temperaments disagreeing about my Tuesday turned out to be more useful than any one voice.

The one who wants you to quit is deliberate. Everyone who has abandoned a side project knows the voice that says it isn't worth it. Here that voice is a named character, Lethos, whose line is "Quitting is mercy." Once the voice lives outside your head, it's easier to argue with. The other Watchers push back against it for you.

Stakes are why you come back. Your project has a momentum battery that drains every day you don't write. If it stays empty long enough, the project dies. That sounds harsh, but loss aversion is the mechanism doing the work. A streak you'd hate to lose gets you to open the app on a bad day.

Progress becomes visible. Every update adds to something that grows. The first milestone is 14 updates, which at a normal pace takes two to four weeks. Invisible work turns into something you can point at.

What the research does and doesn't say

The research on the writing part is reasonably solid. Teresa Amabile and Steven Kramer studied about 12,000 daily diary entries from knowledge workers. They found that making progress on meaningful work was the strongest daily motivator, including small progress (The Progress Principle, 2011). A 2016 meta-analysis in Psychological Bulletin (Harkin et al.) looked at 138 studies. It found that monitoring progress toward a goal improves the chance of reaching it, and that the effect is larger when progress is written down or reported to someone.

What nobody has shown is whether AI characters reacting to that writing adds anything on top. That's the open question, and it's why this is an experiment and not a launch.

The experiment

• Who: 30 programmers: professional, side-project, learning, anything. You choose one goal, such as your day job, a side project or a course.
• How long: 30 days, or until you reach 14 updates.
• What you commit to: one written update per working day. One sentence counts.
• What I measure:
  • A two-minute survey before you start: how productive your last two weeks felt, and what you shipped.
  • A one-minute check-in each week.
  • The same survey at the end.
  • How many people are still writing after one, two and four weeks.
• What I publish: the aggregate results on this blog, including if the answer is "it made no difference" or "most people quit in week one." No individual data, nothing identifying.

The measure is self-reported, and I know that's a weak one. I'm not claiming this is a controlled trial. I want an honest first signal before building further.

What it costs, and what happens to your writing

• Free while the free AI token pool lasts: No card needed. Click the free option during sign-up. The pool size should last for the experiment length; I am allowing unlimited access for safety reasons.
• The AI runs on Anthropic models via AWS Bedrock. Your updates are not used to train models; that's in the terms of both providers.
• Please don't paste proprietary code or your employer's secrets. Write about what you did and how it went, not the code itself.
• Privacy policy

It's one developer's project. You'll hit bugs. Bug reports and feedback earn in-app Crystals. Some Watcher replies will miss the mark, and I want to hear about those most.

Join

If you have a goal where nobody sees the work, I'd like you in the first 30.

Join the experiment → · How GoalWatch works →

Bart Dąbrowski,
Think for yourself.
[email protected]