Nine early-stage projects that hit our radar, real products rather than memecoins. The theme: the fine print. Most of crypto sells you the upside and hides the rest. This week's names put the rest in plain view: who takes the first loss, what sits inside a yield, the most you can lose on a trade, which of thousands of stock tokens you actually hold, who owns the robot and who answers for an AI agent. Follower counts are small on purpose, you are seeing these early. With a link to each.

1/ Akashi (Tranche Vaults)

Most yield vaults pay everyone the same rate and share every loss the same way. Akashi splits one Solana USDC vault into two layers: Junior takes the first loss and earns more, Senior is paid first and earns less. Its Akashi Plus vault, live on mainnet since late October, spreads deposits across OnRe, Hastra, Huma and Kamino. At the time of writing it holds $24,920, with Senior at 8.01% and Junior at 14.09% expected, and a Junior cushion of 25.8%. It won the Superteam Poland Demo Day and was one of 10 teams out of 180 picked for the Crypto World's Fair audit grant.

2/ Kurtosis Labs (Risk Ratings)

One onchain yield can be a token, wrapped, deposited, split and lent again, and each layer adds its own risk. Kurtosis Labs rates that yield layer by layer, so a fund analyst can see what they actually own and explain why it moved. A second tool, Horizon, shows how fast you can get your cash back out. It is built for institutions and works by demo request. The team took second place and $40,000 in the General Track at Arbitrum's Founder House in Singapore, and it is backed by Superteam Singapore.

3/ Vexi (Stock Options)

Leverage on stocks usually comes with a liquidation price. Vexi sells calls and puts on stock tokens on Robinhood Chain instead: as a buyer, the most you can lose is what you paid. On the other side, holders pick the price they would sell a stock like NVDA at and earn interest while they wait. The project spells out the risk too: the pool pays winning buyers, and that can cost more than the premiums it collects. Testnet is live and the mainnet whitelist is open. It placed second and won $6,000 in the Promising Products Track at Arbitrum's Founder House.

4/ SoFinance (Stock LP)

The same stock now trades as several tokens from different issuers. SoFinance turns that mess into income: you pair two tokens of the same stock in one pool and collect a share of the swap fees, without selling the stock. It runs on Solana and Robinhood Chain, and the keys stay with you. Its own data shows the same-stock TSLA pool on Meteora did $1.30M of volume over 7 days, against $10.3K in Raydium's TSLA pool. Fees are real, so are out-of-range and impermanent loss risks.

5/ Darwin (Private Baskets)

Building a portfolio onchain usually means many trades and a wallet anyone can read. Darwin packs a whole strategy of crypto and tokenized stocks into one token: one trade buys the full basket, and you hold it privately in your own wallet on Miden, a zero-knowledge chain. On November 7 it went live inside the Bread wallet. It is in beta, and the account is still tiny.

6/ tokenizedstonks (Data & Analytics)

There is no longer one Tesla token, there are dozens. tokenizedstonks groups every tokenized version of a stock, ETF or index under the real security, across issuers, chains and venues, with the source and time of each figure. On Solana alone it tracks more than 1,700 securities across more than 2,700 tokens. It is a free tracker, not a protocol, and it earns from disclosed referral links.

7/ RoboPay (DePIN Payments)

When a robot delivers your parcel, someone still has to get paid. RoboPay locks the buyer's payment in a Solana escrow, the robot scans the box to prove it arrived, and the money goes to whoever owns that robot's machine NFT on peaq. Sell the NFT, and the next delivery pays the new owner. It won first place in the peaq track of the Crypto World's Fair hackathon. For now it runs on Solana devnet, with a working demo and a public solo founder.

8/ HestiaX (Agentic Finance)

AI agents can move money faster than any team, which is exactly the problem. HestiaX builds agent workflows for energy transition finance, where the agents do the work and humans sign the mandate they must stay inside. Its motto: AI automates, humans mandate. It also ships an XRPL consensus explorer and runs a validator. The team is public and won the Ledger Agentic Finance award at the XRPL hackathon in New York. Still early, with the product at demo stage.

9/ Entity.ID (Onchain Identity)

When an AI agent signs a deal or moves funds, the first question is who is behind it. Entity.ID keeps one public record per company, fund, DAO or AI agent: who owns it and who answers for it. It builds on ERC-8004, the Ethereum standard for agent identity, and the app is already live. The founder spent Devcon 8 in Mumbai putting that exact question to every room.

The pattern this week: the fine print is becoming the product. A vault that says who loses first, a rating that opens every layer of a yield, an option where the worst case is printed upfront, a map of thousands of stock tokens and a public record of who stands behind an AI agent. None of it promises 100x. All of it tells you what you are holding. These are early-stage projects, several still on testnet, on devnet or in beta, so higher upside comes with higher risk. Do your own research, nothing here is financial advice.

Which one are you watching?