What Can You Build With Claude Opus 5.5? The Apps I Built for Myself

This one is a little different. It is not a benchmark and it is not a tutorial. It is a show and tell of what I have built with Opus 5.5 that I am just amazed by.

Keep in mind, these are not the most crazy visual use cases. The apps I am most excited about are the ones that fix my own daily friction and put my own experience to work. I am sharing them in hopes of inspiring you to build something for yourself.

The quick version

| App | What it is | What it took | |---|---|---| | Bricklands | A one for one Minecraft clone reimagined in the Lego universe | One prompt, 19 minutes | | Edge | An NFL and DFS research platform | Roughly 100 prompts | | Sleepers | Tracks beat writers across every NFL team to surface under-the-radar players | Still fine-tuning | | Clearmud HQ | One workspace that holds all my Clear Mud tools | In active development |

I also tested Opus 5.5 on every effort level by cloning Sonic the Hedgehog, one of my all-time favorite childhood games. That has its own video, so I skipped the demo here.

Bricklands: a Lego Minecraft in 19 minutes

I know, I know. These demos are oversaturated. But it actually nailed it.

Here is the prompt I used:

> I want you to build me a Minecraft clone, but reimagined as if we're in the Lego world. I want you to make all the tech stack and creative decisions yourself. Keep in mind that our goal was originally to clone Minecraft one for one, but instead of looking like Minecraft, it should be Lego World.

It did it in 19 minutes.

Everything works. I can collect dirt. I can build crafting tables and weapons. I walked up to something I did not recognize and it turned out to be a pumpkin. There are pigs. There is a snow biome.

It was not perfect. When I went back to single player, my saved worlds from the original session were not there. That's a bummer. But the speed is what I'm amazed by. I handed over every tech stack and creative decision, and 19 minutes later I had a working game.

Edge: a decade of fantasy football in one platform

This is the one I am in love with.

I am a big NFL fan. I have played fantasy football for well over a decade, and I have been doing DFS and sports betting for about ten years. Edge is an accumulation of that decade of knowledge and experience and failures and expertise, pulled from all the sources I used to check separately into one platform.

Keep in mind, Opus 5.5 did not one-shot this. What you see in the video is roughly a hundred prompts. Back and forth, fine-tuning and adding, over and over.

Here is what it gives me:

• Slate view. Highest team totals, DraftKings value leaders, weather watch, and the main slate with color-coded game totals so I know which games to target. I can filter by main slate, early, afternoon, or showdowns.
• Game pages. Where the edges are, unit matchups, and pass and rush volume for every week of the season, so I can see whether a team was pass heavy or balanced.
• Player trends. It does not just show that a snap percentage changed. It explains why, like a teammate who did not play or a rookie whose workload is growing. This took about 20 prompts to get right and accurate, and I back-tested a lot of it.
• WR vs CB matchups. This took about 30 prompts. It makes spots obvious, like a week two game where the top receivers were out and one veteran was left on the field with the depth guys. That is an all-in type of play.
• Player pages. My favorite feature is history versus a team, with a home icon for players returning to their hometown. They buy tickets for friends and family, and they tend to go off. It does not hit 10 out of 10, but I would say 80 to 90% of the time it hits. About two years ago I went all in on Tank Dell going back to Jacksonville at a cheap DraftKings price. He scored 28.5 PPR points.

It also misses, and I showed that too. I added the defense page the night before recording because Pittsburgh at home against a divisional opponent is usually dominant. That day they scored about five fantasy points. The page also jumped ahead to week four because the game had already been played, so I might need to add a toggle in there.

Sleepers: finding the players nobody is talking about

Sleepers catalogs beat writers from every NFL team, plus the major outlets, to surface under-the-radar players. One example from the video: with a top receiver out, it flagged that 27% of the team's targets were open and pointed to the next man up. He did not do well that day, and I said so.

I am still fine-tuning it. The plan is a free launch at sleepers.app.

Clearmud HQ: one place for everything

Here is a confession. I built Canvas and my ideation page in public, and then I stopped using them. I found myself not logging in because it was inconvenient. They lived in separate places and I would forget.

Clearmud HQ fixes that. It is one workspace, so I do not have to keep tabbing into different apps.

• Embed, don't rebuild. Clearmud OS stays on its own web server and port, untouched. HQ pulls each app's left menu into one unified menu.
• Split views. I can put pages from different apps side by side. One split showed some weird functionality on camera, and I left it in.
• Workspaces. My YouTube workspace lines up ideation, script, content calendar, and then Canvas, so I can sit down and start working.
• A Build tab. It spins up a coding session per app, with a model toggle and live and preview tabs. Each change lands on its own branch for me to approve before it goes live. No separate CLI session needed.

I tested the Build tab live. I asked it to bring the ideation app up to the same updated styling I had already applied to the content calendar. I even said please.

What I took away from all this

• Build for yourself first. Pick something you are passionate about and build the tool you wish existed.
• Start from your own expertise. The model built Edge, but I knew which stats and edges mattered.
• Expect iteration on anything real. One prompt gets you a toy. A hard feature costs 20 to 30 prompts, and a full platform costs around a hundred.
• Hand over the decisions on quick builds. When speed matters more than control, let the model pick the tech stack.
• Back-test before you trust it. Do not take the output at face value.
• Centralize your tools. If you keep forgetting an app you built, the problem is access.
• Ship changes on a branch with a preview. Review it next to the live version, then approve.

Unlimited, untapped potential is at our fingertips here. Opus 5.5 is living up to the hype.

I'm not an AI expert. I'm building in public and sharing what actually works. I just wanted to show off what I've built for me, what I'm genuinely excited about, to hopefully inspire you to do the same for whatever it is you're passionate about.

If you want a video tailored to you, drop a comment on the video with who you are, what you do, your target audience, and your question. I will add it to my queue.

Watch the full show and tell: https://www.youtube.com/watch?v=Pj7MxLyivAA