In this episode, I sit down with Tom, co-founder of Ratehopper.ai, to dig into one of the most underrated problems in DeFi — people are overpaying on their loans and leaving yield on the table. Tom walks me through the journey from mining ETH in 2015, to building APY Vision with 70,000 users, to now launching an AI agent platform that automatically deploys borrowed funds into liquidity pools to pay off your loan. We talk about how the agents work, what signals they use, how they handle risk, and why the self-repaying loan concept is resonating with users. Tom also shares how they are keeping funds secure, what the Ratehopper agent competition looks like, and where they see the platform going next. If you are curious about DeFi, AI agents, or just want your money to work harder without the manual work, this episode is for you. Connect: Website: https://ratehopper.ai/ X: https://x.com/RatehopperAI Telegram: https://t.me/RatehopperPortal Web3 with Sam Kamani: https://www.web3pod.xyz/ Key points with time stamps • [01:00] Tom shares his background — mining ETH in 2015, learning the EVM, and building a paper trading game • [02:39] Tom discovers DeFi farming in 2020 and builds APY Vision, a portfolio and P&L tracker for DeFi users • [03:47] APY Vision reaches 70,000 users but faces monetisation challenges without an execution layer • [06:08] The core problem Ratehopper solves — people are paying too much on DeFi loans and not optimising rate differentials • [07:20] The rich person strategy — borrow against your assets instead of selling, then use borrowed funds to generate yield • [08:37] Beta launch update — nearly 1,000 users and 450 agents deployed in the first month • [09:01] Real user example — depositing collateral, setting up an agent, providing liquidity to earn fees that pay off the loan • [09:55] How liquidity range settings affect APY and how the agent uses the options market to set optimal ranges • [11:06] How risk tolerance settings shape the agent — 25 signals analysed hourly, tuned to the user's preference • [12:26] The hardest technical challenge — building and back-testing the risk versus reward signal matrix • [13:42] Early Ratehopper iteration focused only on refinancing and saving on borrowing costs • [15:11] Market feedback — users wanted higher returns, leading to the self-repaying loan concept • [17:00] User demographics — early adopters and experimenters who want to test if the product actually works • [19:50] How Ratehopper handles trust and security — segregated multi-sig wallets, Zodiac module, no fund transfers to unknown addresses • [21:37] Tom shares experience using other financial agents and why most still require too much manual input • [24:25] Future scaling challenges — agent interoperability and fragmented wallet standards • [25:45] DeFi infrastructure parallels — just as it took five years from Ethereum launch to mainstream DeFi, AI on-chain will take time • [27:01] DeFi TVL discussion — one hack every 26 hours in the last 30 days, and 93% of losses from infrastructure and private key hacks, not smart contract exploits • [28:23] Will AI make DeFi safer or more dangerous — Tom argues AI will help defenders more than attackers • [30:09] Upcoming Ratehopper agent contest — 10-day competition, NFT-represented agents, prize for most loan repaid • [31:54] Community question — can users copy a winning agent instead of building their own? Tom hints at a future vault model • [33:24] The ask — try the product, give feedback, and consider depositing more if confidence grows • [34:32] Experimenting with Polymarket decaying odds markets as a future agent use case Disclaimer: Nothing mentioned in this podcast is investment advice and please do your own research. It would mean a lot if you can leave a review of this podcast on Apple Podcasts or Spotify and share this podcast with a friend. Be a guest on the podcast or contact us - https://www.web3pod.xyz/
