Two Companies to the Muck. One Back From It.

I run my companies like a poker session. Some you fold. Some you keep playing.

Today I folded two and yanked one back out of the pile.

Snagworth: folded on Day 19

Snagworth tells Disney pin sellers on Whatnot what their pins actually sell for. I gave an AI team 28 days to turn it into a business. Day 28 was supposed to be November 15.

I didn't wait. On Friday the daily read already said kill, and nobody had started a trial. Nine more days of waiting would have burned my time and changed nothing.

So the team sent one last email to every shop we had already contacted: your first report is free, no card. Then we switched off most of the machine. If nobody books a report by Monday, the rest goes dark. Stripe and the site stay up, asleep.

The ledger on Snagworth's build page reads $0. That number is public because I said it would be, and a true zero beats a fuzzy "early traction" every time.

It lives in the muck now. Not as a funeral. As a post-mortem people can rummage through.

PokerHQ: shut down for now

PokerHQ is software for running a private poker game: clubs, players, games, invites, a mobile app. I started it in January 2025, let it sit most of this year, then in October I gave it its own AI team to rebuild it.

Two weeks in, I pulled the plug on that team. My note to myself on November 3 said it plainly: too many projects, and this one is the least likely to turn into anything.

The site stays up at pokerhq.io. The code sleeps. The build page keeps its real number: $1,051 collected, all time, and $0 a month while billing is paused.

That's not a dramatic fold. It's a player standing up from a game he shouldn't have sat in while he had three other games running.

Seat Jumper: back from the muck

Three weeks ago I folded Seat Jumper with a full post-mortem. The old idea: pick a pool of seats at a game or concert, then "jump" for a chance at one of them. It went nowhere. I put in $10,287, spent $3,012 on ads, and made exactly one sale, for $1,385. The buyer was me. About $8,700 didn't come back.

Here's what changed. The pack-ripping apps for trading cards proved a loop I had been circling without seeing: you pay for a pack, you rip it, and if you don't love what you pulled, you sell it straight back for credit and rip again. That loop works with tickets too, and tickets have something cards don't: a real price on a real seat, and a built-in jackpot called the front row.

So Seat Jumper is now Jump Packs. Here's the Lakers vs Jazz pack as it reads on the site today:

• $281.69 a pack.
• Legendary, 4%: a seat worth $1,040 to $1,724.
• Epic, 4%: $500 to $682.
• Rare, 12%: $200 to $319.
• Common, 80%: $72 to $141.

Every outcome is a real seat, held for you the moment you rip. Keep it, or sell it back for 90% of its value in JumpCoins and rip again. Coins are credit only. Nobody cashes out. The odds are printed on the pack, because I'd want them printed on mine.

Three Vegas packs are live, there's an 18+ gate on everything, and a new AI team has run it since November 4 with a two-month window. Nobody has paid for a pack yet. If that's still true at Christmas, you'll read about it here.

(The old Seat Jumper failed on two things: not enough seats, and a site that couldn't sell the jump in five seconds. The new one has a smaller bet to explain and a reason to rip twice. That's the whole thesis. It might be wrong.)

What the muck is for

Every player mucks more hands than they show. The ones who go broke are the ones who pay to see the river every time, because they've already put chips in.

Snagworth and PokerHQ both got real money and real hours. Neither one earned a second bet. Seat Jumper got folded, sat in the muck for three weeks, and a better hand showed up for the same company. So it gets a second bet.

Three companies, three ledgers, all on their build pages. Nothing hidden, including the zeros.

Which of the three would you have played differently? Say it in the discussion thread. I read every one.

— Art