Cross-Chain DeFi,
Operated by AI Swarms
A unified DeFi hub for 64+ chains with an AI agent layer that turns natural language into multi-chain strategy.
"Cross-chain DeFi is the most fragmented user experience. The job was to make 64 chains feel like one product, then put an AI layer on top so users could stop thinking about chains at all."
01 — The Context
DeFi is supposed to be open. In practice, it is fragmented across 60+ chains, each with its own bridges, DEXes, staking yields, gas tokens, and edge cases. A user holding ETH on Arbitrum who wants exposure to a Solana liquidity pool ends up bridging through three protocols, paying gas in three currencies, and praying nothing slips.
Chainbased came to us with a wide ambition. Build a single platform that absorbs all of that complexity. 64+ blockchain networks. 13+ bridges. 36+ staking providers. 12+ aggregator DEXes. FIAT on-ramps. Real-world assets. And on top of all of that, an AI agent layer — the Chainbased Agent Nexus.
Networks
64+ blockchains unified
Bridges
13+ bridge protocols
Staking
36+ staking providers
DEXes
12+ aggregator DEXes
02 — The Bet
The chain becomes invisible.
The user states an intent and the platform figures out the route.
— 01
Chain as implementation detail
The default move in cross-chain DeFi is the network selector. We bet the opposite. The interface surfaces what the user actually cares about — the outcome and the cost, not the protocol underneath. That decision made the AI agent layer not a feature but the natural top of the funnel.
— 02
A multi-product platform, not a hub
Most DeFi hubs ship as a swap interface and call it a hub. We treated Chainbased as a multi-product platform from day one. Swap, Bridge, Stake, RWA, FIAT on-ramp, Pools, Gas Refill, and the AI agent layer all share a coherent design language and navigation model without collapsing into a generic dashboard.
03 — My Role
Product Designer
Design System
Owned the design system that holds the cross-chain DeFi surface and the AI agent layer together.
Intent Dashboard
Designed the intent-centric dashboard where users type goals in natural language and the platform routes strategy.
AI Adapter Studio
Designed the no-code tool for building neural adapters — a natural language configuration interface for on-chain logic.
Swap & Bridge Flows
Abstracted multi-hop, multi-chain routes into a single transaction surface. Power users expand; most users never need to.
Swarm Analytics
Designed the swarm performance analytics surface — each agent role visible and traceable inside a running strategy.
$BASD Visual Identity
Shaped the visual identity for $BASD and the broader Chainbased ecosystem.
Seam Decisions
The hardest design decisions sat at the seams — where swap meets bridge, where swarm meets user, where card meets chain.
04 — The Build
The AI Agent Nexus
Three swarm roles, each with its own UI treatment. Users see what each agent is doing and why — every action traceable, every decision explained.
Liquidity Scout
Monitors undervalued pools across all connected chains. Surfaces yield opportunities before the crowd reaches them.
Sample Task
Detect Arbitrum USDC pool with 12% APY spike vs. 7-day average.
Alert issued, swap path pre-computed, waiting for confirm.
Risk Assessor
Scores every strategy on impermanent loss, MEV exposure, and smart contract vulnerability before capital moves.
Sample Task
Evaluate LP position in new ETH/MATIC Uniswap V3 range.
Risk score 6.2/10. IL probability 34% at ±8% price band.
Executor
Deploys capital and shows the execution trail step by step. Every on-chain action is surfaced with its reason and gas cost.
Sample Task
Rebalance portfolio to 60% blue chips, route via Chainlink CCIP.
4 transactions. 2.3s total. $1.82 all-in gas. Complete.
The Intent-Centric Dashboard
Users type goals in natural language: "hedge my ETH against a market crash", "find me the best stablecoin yield with low IL risk", "rebalance my portfolio to 60% blue chips". The dashboard parses the intent, recommends a swarm, and shows the projected outcome before capital commits.
The interface is designed around the user's goal, not the protocol used to reach it. Chain selection, bridge routing, and DEX aggregation are absorbed into a single surface. The user sees a recommended path and a cost, not a routing diagram.
The Adapter Studio
A no-code tool for building custom AI adapters. Users describe what they want and the studio compiles it into a runnable adapter. Pre-built adapters live in a marketplace priced in $CAN tokens.
Swap & Bridge as One Surface
A single transaction even when the route runs through three bridges and four DEXes. Final outcome, all-in cost, time estimate, and protocols as collapsed metadata. Power users expand the route. Most users never need to.
"The default move in cross-chain DeFi UX is the network selector. We bet the opposite. The chain becomes invisible. The user states an intent and the platform figures out the route."
05 — Tech & Stack
06 — The Result
150K+
Wallets connected on testnet
100K+
Community members
$50M+
Testnet transaction volume
2M+
Transactions on testnet
10K+
TPS on testnet
64+
Blockchain networks live
$3M
SAFT round raised
Q1 2025
Token Generation Event
Numbers below the visible line: a product that compresses an entire category of DeFi tooling into a single navigation model, and an AI layer that gives curious users a real entry point that doesn't require them to learn the chain map first.
07 — What It Unlocked
Chainbased is the project where we proved that cross-chain UX is not a navigation problem. It is an intent problem. Once the user states what they want, the chain becomes an implementation detail.
More than anything, this is the project that turned the AI agent service line at Luvon from a future bet into a production discipline. Every cross-chain product we have designed since has inherited this thesis.
Intent-First UX Thesis
Every cross-chain product we have designed since Chainbased has inherited the intent-first model. Chain as implementation detail is now a studio principle.
Adapter Studio Pattern
Natural language as a configuration interface for on-chain logic has become a recurring play. The pattern gets cited most often by founders who reference this work.
AI Agent Discipline
Chainbased turned the AI agent service line at Luvon from a future bet into a production discipline. Swarm visibility and agent-role UX are now established patterns.
"Two products stacked on each other. The cross-chain DeFi surface and the AI agent layer. Both had to feel like one."
"Once the user states what they want, the chain becomes an implementation detail."
Start a Project
Want to build with intent?
Let's talk about your product — the real constraints, the real goals, and the cleanest path between them.
Start a Conversation