Verified usage · updated with the leaderboard · as of Oct 28, 2026

AI trading apps on Hyperliquid, by real usage

Senpi, Minara, DXAP, Wayfinder and other apps that trade for you with AI, compared on what actually happened: how many people used them, how much they traded and whether those users made money, read from every fill Hyperliquid published between Sep 28, 2026 and Oct 27, 2026.

8AI trading apps tracked
1,561users in 30 days
$24.2Mtraded through their builder codes
6 of 8apps whose users lost money overall

All apps compared

Sorted by volume; this is not a ranking of skill. Users' realized PnL is the sum of closed PnL on the app's fills, before exchange fees; open positions aren't included.

AppUsersVolume, 30 daysUsers' realized PnLUsers in profitMedian user
TrueNorth
not clearly stated on its site
283$11.4M−$1,36151%$0.02
Senpi
autonomous strategies and manual or copy trades
831$5.9M−$4,71440%−$0.38
DXAP
autonomous agent, steered by chat
233$3.6M$24k34%−$1.28
Wayfinder
agent trades on request; automated strategies
43$1.4M−$1,08042%−$0.04
Midas
copilot: AI suggests, the user taps to trade
80$1.1M−$4,06759%$0.38
Katoshi
bots from alerts, webhooks or AI agents, plus a manual terminal
16$632k$63238%−$1.56
Minara AI
autopilot, AI copilot and manual orders
73$254k−$23747%−$0.21
Starchild
copilot: orders from chat instructions
2$4,327−$10%−$0.85

Each app's numbers

TrueNorth

283 users traded $11.4M through TrueNorth's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was −$1,361; 51% of users with closed trades were in profit and the median user made $0.02. No builder fee was charged on these fills.

How it trades, by its own description: not clearly stated on its site. Source.

Builder address 0x30599f69164b854a2c10ada95e5cd219b72d5216, from DeFiLlama's open-source factory/hyperliquid.ts adapter.

Senpi

831 users traded $5.9M through Senpi's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was −$4,714; 40% of users with closed trades were in profit and the median user made −$0.38. The app earned $2,522 in builder fees on these fills.

How it trades, by its own description: autonomous strategies and manual or copy trades. Source.

Builder address 0x1368f4311db5807f7c7924d736adaeb83e47bafe, from DeFiLlama's open-source factory/hyperliquid.ts adapter.

DXAP

233 users traded $3.6M through DXAP's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was $24k; 34% of users with closed trades were in profit and the median user made −$1.28. The app earned $894 in builder fees on these fills.

How it trades, by its own description: autonomous agent, steered by chat. Source.

Builder address 0x5803667556f289ae854c914e59a2395b75c8aa24, from its own docs or code.

Wayfinder

43 users traded $1.4M through Wayfinder's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was −$1,080; 42% of users with closed trades were in profit and the median user made −$0.04. The app earned $644 in builder fees on these fills.

How it trades, by its own description: agent trades on request; automated strategies. Source.

Builder address 0xaa1d89f333857ed78f8434cc4f896a9293efe65c, from its own docs or code.

Midas

80 users traded $1.1M through Midas's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was −$4,067; 59% of users with closed trades were in profit and the median user made $0.38. The app earned $540 in builder fees on these fills.

How it trades, by its own description: copilot: AI suggests, the user taps to trade. Source.

Builder address 0x990a91cb54c6db7037d3f8105e848c635d4d909f, from DeFiLlama's open-source factory/hyperliquid.ts adapter.

Katoshi

16 users traded $632k through Katoshi's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was $632; 38% of users with closed trades were in profit and the median user made −$1.56. The app earned $156 in builder fees on these fills.

How it trades, by its own description: bots from alerts, webhooks or AI agents, plus a manual terminal. Source.

Builder address 0x274e3cdb7bdc4805f41a07e3348243ba3e7e5b72, from DeFiLlama's open-source factory/hyperliquid.ts adapter.

Minara AI

73 users traded $254k through Minara AI's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was −$237; 47% of users with closed trades were in profit and the median user made −$0.21. The app earned $47 in builder fees on these fills.

How it trades, by its own description: autopilot, AI copilot and manual orders. Source.

Builder address 0x5a3bc60b0a99a7f4fbf0d15554fa5fe88e7628c2, from DeFiLlama's open-source dexs/minaraai-perps.ts adapter.

Starchild

2 users traded $4,327 through Starchild's builder code between Sep 28, 2026 and Oct 27, 2026. Their realized PnL on those fills was −$1; 0% of users with closed trades were in profit and the median user made −$0.85. The app earned $1 in builder fees on these fills.

How it trades, by its own description: copilot: orders from chat instructions. Source.

Builder address 0x2c5320f40305ffc933385c6dcec5493fba7b98b8, from its own docs or code.

Not included: Signalview (operator and AI claim not confirmed); AlphaPilot (not confirmed to be the same app as the AI product of that name); Sendai (matched by name only). An app is listed only when its own site or code says it runs AI agents or an AI copilot for users, and its builder address comes from its own docs or code or DeFiLlama's open-source adapters.

Read these numbers right

Only tagged fills count

Trades sent another way, or on plans that don't add the builder code, are missed, so an app's real usage can be higher.

Users, not the AI

Apps mix autopilot, copilot and manual orders. These numbers show how users did with the app, not how the AI alone did.

Before exchange fees

Realized PnL comes from closing fills and is before Hyperliquid's trading fees and funding, so the real result is a little worse.

30 days is short

One month says little about skill. Watch the trend on the leaderboard, which updates with every data refresh.

FAQ

What is a Hyperliquid builder code?

A builder code is an address an app attaches to the orders it sends to Hyperliquid so it can charge a small fee. Hyperliquid publishes every fill that carried a builder code, which is how we can count each app's users, volume and results.

Do people make money with AI trading apps on Hyperliquid?

Over the last 30 days, users of 6 of the 8 apps we track lost money overall. Their combined realized PnL was $13k on $24.2M traded, and DXAP's users alone made $24k. In most apps fewer than half the users were in profit.

Does a profit or loss here mean the AI made that trade?

Not always. Several of these apps mix autopilot, copilot suggestions and manual orders, and the public fills don't say which was which. The numbers show how the app's users did, not how its AI did.

Why is an app I use missing?

Either we couldn't confirm its builder address from its own docs, code or DeFiLlama's open-source adapters, or its own site doesn't say it runs AI for users. Send it to us with the source.

Open data: AI trading apps on Hyperliquid (JSON). Also see do AI trading bots make money? and the Hyperliquid listing. No affiliate links on this page. Not financial advice. Hyperliquid's terms restrict US and Ontario (Canada) users.

Get the weekly agent risk update

New Agent Risk Scores, incidents and safety changes for the wallets, exchanges and brokers AI agents use. One email a week, unsubscribe anytime.

No spam. Unsubscribe anytime. Privacy