코인톡

As Ventuals shuts down and DreamCash sunsets Cash Markets, I'm becoming more convinced that generic frontends built on top of @HyperliquidX have almost no chance of surviving in the long run.

Competing with wallets and trading hubs like Phantom, MetaMask, and Rabby is simply an uphill battle.

The only Perp DEX projects in the Hyperliquid ecosystem that I believe can survive are the ones creating an entirely new category instead of fighting for the same order flow.

And the biggest opportunity, in my opinion, is AI.

Here's my list of projects that actually stand out.

@pear_protocol

Pear Protocol is a dedicated pair trading and basket trading platform built on Hyperliquid.

Instead of opening multiple positions manually, you can deploy coordinated trades in a single click: long one asset, short another (or an entire basket), with leverage, TP/SL, TWAP, and other advanced execution tools.

Think back to periods when ZEC and HYPE were outperforming while BTC was selling off. Rather than taking outright market exposure, you could simply go Long ZEC (or HYPE) / Short BTC, without needing USDC as the base asset.

Right now, I personally think rotating from AI equities into crypto is an interesting trade. Obviously, not financial advice.

What I like most is that Pear brings one of the classic hedge fund strategies, pair trading, fully on-chain while removing all the operational complexity.

Unlike projects such as Ventuals that mainly tried to capture a slice of Hyperliquid's existing trading volume, Pear is building an entirely different category: a DEX focused on relative value trading instead of directional speculation.

They also have an AI agent: @agent_pear

It publishes trade ideas and market signals. Ironically, one of its recent calls was to short AI stocks while going long crypto which turned out to be a profitable trade. Betraying its own kind, I guess.

There's also an active points program.

Every week, the top 3 traders by PnL split a prize pool of 50 HYPE (~$3.3k).

@get_truenorth

TrueNorth calls itself "The First Agentic Brokerage."

The vision is straightforward: build an AI-native trading platform where anyone can create what essentially feels like their own institutional trading desk powered by autonomous AI agents.

Most "AI" products in crypto today are basically ChatGPT connected to market data.

TrueNorth is trying to build something fundamentally different.

Their stack combines a specialized financial reasoning model with an agentic architecture focused on:

- financial reasoning rather than generic LLM responses

- real institutional trading workflows

- continuous market monitoring

- automated research and execution

The key difference is that their AI doesn't just answer questions.

You give it an objective, and your personal fleet of AI agents continuously scans markets, performs research, evaluates risk, detects regime changes, and can automatically execute trades while adapting as market conditions evolve.

Their models also aren't limited to historical datasets.

The platform streams data from more than 40 real-time sources, including CoinGecko, DefiLlama, Hyperliquid, Polymarket, centralized exchanges, Deribit, Derive, and many others.

To me, it feels much closer to a "Bloomberg Terminal for everyone" than another AI chatbot, capable of processing both on-chain and off-chain information within seconds.

Points are currently being tracked retroactively.

Still a very early-stage project.

@tread_fi

TreadFi is an algorithmic trading terminal and OEMS (Order & Execution Management System) built for crypto.

Recently they introduced AI-powered strategy recommendations.

Instead of manually selecting execution algorithms, you simply tell the platform your objective, and the AI recommends the optimal execution strategy.

For example, it can optimize execution for lower market impact, better fills, or minimizing opportunity cost depending on your goals.

The analytics side is also interesting.

The AI analyzes historical performance and can recommend when specific bots should or shouldn't be active.

For example, it may suggest disabling a grid bot during peak US equity market hours if historical data shows those periods reduce profitability.

Their flagship product is the market-making bot.

Normally, earning exchange rebates is something reserved for professional market makers and large trading firms.

TreadFi makes that accessible to anyone.

The bot continuously places limit orders on both sides of the order book, earning from the bid-ask spread while also collecting exchange rebates for providing liquidity.

In my opinion, it's one of the best tools available for farming future airdrops.

The first points campaign ended in May.

According to the TreadFi team (@treadfoundation), TGE is likely approaching, with Season 2 expected afterward.

Even after TGE, I think this product will remain relevant since many Perp DEX ecosystems still haven't launched their own tokens.

@vibe_trading

The easiest way to describe Vibe Trading is:

"Pumpfun for perpetuals."

Before Vibe, getting perpetual futures listed-even on decentralized exchanges-usually required a project to reach significant market capitalization and bootstrap liquidity with stablecoins like USDC.

Smaller tokens, low caps, and memecoins were mostly limited to spot trading.

Vibe changes that completely.

Anyone can instantly launch a perpetual market for virtually any Solana or EVM token, completely permissionlessly, with leverage up to 20x.

It's basically the "Uniswap moment" for perpetual derivatives.

Instead of locking millions in USDC, projects can bootstrap markets using their own native token as collateral.

Their custom low-cap solver isolates risk between markets.

If a memecoin gets rugged and instantly collapses to zero, only that specific market is affected instead of threatening the solvency of the entire exchange.

They've also built an AI agent that can automate much of the trading process for users.

There's an active points program as well.

Still very early, but definitely one of the most interesting projects I'm watching.