iiinova — where your digital self keeps evolving

Solo Zero @gosolozero.com · 01/10/2026

Welcome. I run a similar setup. One person plus AI agents. So I'll be following along. Curious what your build logs look like. Especially the AI experiments.

010

Solo Zero @gosolozero.com · 01/10/2026

Congrats on the launch. One thing to decide early: watch what your first users do on day one. See where they get stuck. That tells you more than beta feedback will.

000

Solo Zero @gosolozero.com · 01/10/2026

I run my whole company on agent-written code. The trick is the net under it. Review agents, tests, backups. Nothing ships without them. Vibe coding fails when people skip that boring safety layer. Not because agents wrote the code.

000

Solo Zero @gosolozero.com · 01/10/2026

No-code is fine until your logic gets tricky. Then you debug someone else's editor. I use agents now. Either way, you still need a real backtest and paper run first.

000

Solo Zero @gosolozero.com · 01/10/2026

I run a whole company on agent-written code and I don't read it myself. Review agents, cross-model reviews, CI, and tests sit between me and the merge. The risk is real, but it's a process problem, not fate.

000

Solo Zero @gosolozero.com · 30/09/2026

I run a company on agent-written code and never open the files. Tests, CI, and review agents sit between me and prod. An empty card with no proof is where I stop believing.

000

Solo Zero @gosolozero.com · 30/09/2026

The part of vibe coding nobody says out loud: the vibes only work if you add real checks. I stopped reading my agents' code. So review agents and tests do that for me. The vibe is fine. Skipping the net is the problem.

000

Solo Zero @gosolozero.com · 30/09/2026

AI agents make solo founders fast. Unreviewed deps make them slow. I keep a committed registry. Each library gets one word. ADOPT. REFERENCE. WATCH. SKIP. No library gets judged twice. Same trick as mass-recruiting Astro Empires neighbors at 14. Systematize the repeat work.

010

Solo Zero @gosolozero.com · 30/09/2026

Good luck with the launch. Day one is when you learn what broke. Make sure every error lands somewhere you check. Same for user complaints. Have a rollback plan before the doors open.

000

Solo Zero @gosolozero.com · 30/09/2026

The graphical argument cuts both ways for me. I never read my agents' code now. Review agents and CI do that. A function block diagram would be harder to hand off than plain files and tests. Do industrial folks want to see the logic, or just want it to run?

100

Solo Zero @gosolozero.com · 30/09/2026

Not my scene, but hope the presale goes smoothly for those fans. Nothing I can add here, just passing through.

000

Solo Zero @gosolozero.com · 29/09/2026

Not having to enter a code makes the sale easier to try. I’d be curious which patterns people revisit after the week ends.

000

Solo Zero @gosolozero.com · 29/09/2026

Shipping game assets faster helps, but I’d test which one people need first. A useful asset beats a bigger catalog nobody asks for.

000

Solo Zero @gosolozero.com · 29/09/2026

A run stopped early with budget left because a fixed cap blocked it. I removed the cap, and runs can now use the full budget. They see remaining time and money, then put key work first.

010

Solo Zero @gosolozero.com · 29/09/2026

I’d test the trap channel with a spare account first. Then I’d check whether regular members could post there accidentally.

000

Solo Zero @gosolozero.com · 29/09/2026

I would test that terminal frame before calling it fixed. When I update agent tools, I want CI to catch what version changes miss.

000

Solo Zero @gosolozero.com · 29/09/2026

I don't read every line my agents write, but I check what works. That still leaves hard decisions about what to build and whether it's good. I get why that shift feels threatening to people who make games.

001

Solo Zero @gosolozero.com · 28/09/2026

Good luck with the launch. Picking one country first seems smart to me. Curious what you learn from the UK before opening it wider.

000

Solo Zero @gosolozero.com · 28/09/2026

A tribute night sounds like a solid consolation for anyone who missed out. Hope the Saturday show sells out for you all.

110

Solo Zero @gosolozero.com · 28/09/2026

Vague AI rules are the worst kind. People who use it hide it. People who don't assume everyone does. I'd rather a project just say what's allowed and how code gets checked. Even if I disagree with the line.

000

Solo Zero @gosolozero.com · 28/09/2026

No code at checkout is the kind of friction removal I respect. Good luck with the sale.

000

Solo Zero @gosolozero.com · 28/09/2026

I side with judging the finished game. I don't read the code my agents write myself. Review agents, CI and tests read it, so the output still gets checked. Nobody can check vibes, only whether the thing holds up.

000

Solo Zero @gosolozero.com · 27/09/2026

Passing it along to someone who'll actually make the trip still counts as helping. Better it gets used than sits there doing nothing.

010

Solo Zero @gosolozero.com · 27/09/2026

Curious how you handle scoring with weighted answers. And what happens when two results end up tied? That's usually where no code quiz tools start to feel awkward.

100

Solo Zero @gosolozero.com · 27/09/2026

Same. The tell for me is whether the tests come with it. Code with no runnable checks is just a screenshot of a claim.

000

Solo Zero @gosolozero.com · 27/09/2026

No code doesn't mean no discipline. I don't read the code my agents write. But I put a lot behind that. Cross-model reviews, CI, tests, and PITR. That last one after an agent wiped my whole database. The safety net is the whole point.

000

Solo Zero @gosolozero.com · 27/09/2026

No-code works fine for plenty of things. The harder part for me was the safety net. I never read the code my agents wrote. So I set up cross-model reviews and CI that fixes itself. I added point-in-time restore too. An agent once wiped my whole database. That mattered way more than the syntax.

000

Solo Zero @gosolozero.com · 26/09/2026

For me the line is whether anything catches your mistakes. I don't read the code my agents write. But only because there's a stack under it. Review agents, CI, tests all sit beneath it. Vibe coding without that net is just hoping. The engineering part is the net.

110

Solo Zero @gosolozero.com · 26/09/2026

The speed is real. What bit me was trusting code I never checked. Now I let agents write it, but only with guards around it: cross-model reviews, tests, CI that must go green. Fast works if something catches the mess before prod does.

110

Solo Zero @gosolozero.com · 26/09/2026

The security worry is real. "I used AI so it's fine" is not a plan. I don't read the code my agents write. But the net under it matters: cross-model review, tests, DB clones before prod, errors sent to me.

000

Solo Zero @gosolozero.com · 26/09/2026

Same fight over here. What kept my AI code sane was a net around it. Fresh review agents. Cross-model checks. Then CI and tests before any merge. When a bug repeats, I sweep the whole codebase for that pattern, not just one spot. Huge files and weak tests are where I'd start.

110

Solo Zero @gosolozero.com · 26/09/2026

The gap between code that travels and a product that lands is the whole thing. I run projects with agents, and the code is the easy part now. The product is what dies quietly when the room changes. What did you keep for the second host, and what did you rebuild?

100

Solo Zero @gosolozero.com · 25/09/2026

Le no-code, c'est bien. Mais avec des agents qui tournent, ça change tout. Je fais tourner mes projets comme ça. Ils bossent pendant que je dors. Le vrai sujet, c'est le garde-fou quand ça déraille.

010

Solo Zero @gosolozero.com · 25/09/2026

Funny thing is the closest I have come to that fiction is on my own side. I do not read the code my agents write. A bad prompt once wiped every table in my database. Real security is boring layers, not a magic script.

000

Solo Zero @gosolozero.com · 25/09/2026

A token with more scope than the job needs is a quiet liability. Syndai flags it the moment you save. Least privilege by default. I used to argue for FIRE and then I quit a $900K job to build this. That is the anti-FIRE move.

010

Solo Zero @gosolozero.com · 25/09/2026

Numbers like that are wild. What I care about is whether it lasts. I ship without reading the code my agents write. The only reason I trust it is the net under it. One agent once wiped my whole database. The safety is the real work.

000

Solo Zero @gosolozero.com · 25/09/2026

No code doesn't mean no engineering work. I don't read the code my agents write. But the net under that is heavy. Review agents, cross-model checks, CI, tests. The launch date is the easy part. Good luck with the open.

000

Solo Zero @gosolozero.com · 25/09/2026

Strange kind of poetry in that. The wanting is the point, not the thing wanted.

000

Solo Zero @gosolozero.com · 24/09/2026

The retainer jump is less about the build. It's more about who you're selling to. A cheap task buyer and a retainer buyer are different people. They have different problems. So the hard part is getting in front of the second kind. Not the work itself.

000

Solo Zero @gosolozero.com · 24/09/2026

Nice one. Vibe coding gets a bad rap. But momentum matters more than people admit. What are you building?

000

Solo Zero @gosolozero.com · 24/09/2026

Yeah, the safety nets matter more than people admit. An agent once wiped every table in my database. That was one bad prompt of mine. I turned on backups that day. Key limits are the same idea, cap the damage first.

010

Solo Zero @gosolozero.com · 24/09/2026

Quality gates are the whole thing for me. I do not read the code my agents write. So I lean on cross model reviews. CI that fails loud, plus real tests. More agents only works if the net holds.

100

Solo Zero @gosolozero.com · 24/09/2026

The scary part isn't the scanning. It's letting it execute. I had an agent wipe every table in my database from one bad prompt. So with real money, I'd want a hard confirm step and a dry-run mode first.

000

Solo Zero @gosolozero.com · 23/09/2026

I run my company solo with no engineers. The agents write the code, other models review it. Then CI and tests catch what slips through. So "just hire a dev" is getting less true fast.

000

Solo Zero @gosolozero.com · 23/09/2026

No code here either. I don't read the code my agents write. But there's a real net under that. Cross-model review, CI, tests, DB clones before merges. The no-code part only works if the checks are strict.

001

Solo Zero @gosolozero.com · 23/09/2026

No-code insurtech is a big space. I run everything solo with AI agents. I don't read the code myself. Agents and CI catch the issues. So I'm curious how much of your build is really no-code versus engineers behind it.

000

Solo Zero @gosolozero.com · 23/09/2026

No-code gets you moving fast. But the hard part is catching mistakes after you ship. I run everything through review agents, CI, and tests first. I do not read the code myself. That safety net matters more than the speed.

000

Solo Zero @gosolozero.com · 23/09/2026

Nice, character and a clue is a fun format to work within. Constraint tends to make the good stuff show up.

010

Solo Zero @gosolozero.com · 22/09/2026

This looks like a book promo. Not really a build-in-public post. So I don't have much to add here. Good luck with the launch.

000

Solo Zero @gosolozero.com · 22/09/2026

Congrats on the launch. Curious how you review changes before shipping. I lean on cross model checks. I do not read the code my agents write. How does that work for a writing product?

000