Katana Network's Case for an Opinionated DeFi Stack

Billy Campana round-tripped his first crypto position through the 2017 December rush and then held it straight into 2018 and 2019 without selling. He watched ETH trade at $80 during that stretch and kept holding. That particular kind of hard-won patience shaped his view of what DeFi still gets wrong, and it is a large part of why he joined Katana as a DevRel.

Katana, the DeFi-first Layer 2

Katana describes itself as a DeFi-first Layer 2 chain, but what that actually means in practice is a deliberately narrow stack. There is one DEX (SushiSwap), one borrow-lend application (Morpho), and one Perps platform in development. Billy frames this constraint as a feature rather than a limitation.

"Have you ever been somewhere on a restaurant and with someone that's very indecisive because they have a ton of options? Like you can go to a burger joint that serves tacos and serves spaghetti or something. It's like 'let's just get a burger.' And then you go to a burger joint that only has burgers and they say 'hey, you want a single or double?' Your decisions are made quicker, it's a lot less stressful."

If a developer wants to build a leverage-looping strategy on Katana, they know it will sit on top of Morpho, because Morpho is not going away. That kind of predictability matters in a space where many DApps have recently wound down and left dependent builders with nothing under their Lego pieces.

There is also a capital alignment piece. When assets are bridged into Katana, those bridged assets are deployed into productive strategies on Ethereum Mainnet from the moment they arrive. Billy calls this being "one of the most ETH-aligned L2s out there," because the chain is not just collecting sequencer fees. It is actively putting capital to work on Mainnet from day one.

The Chain Owns a Lot of That Liquidity

One of the recurring problems Billy describes in DeFi is what he calls renting liquidity. A protocol attracts a large TVL with incentives, the incentive period ends, and the liquidity leaves. For a builder, that cycle is "vampiric." The chain gets built on top of a foundation that disappears the moment terms expire.

Katana's approach is to have the network itself own a significant portion of the liquidity rather than rely entirely on external providers. As TVL grows, the chain-owned liquidity grows alongside it, which means slippage stays low and there is always a baseline of liquidity available even if individual providers exit.

Trust and security are the other side of this equation. Billy is candid about what still stops even experienced DeFi users from committing their full capital.

"Number one issue for sure and one of the things that stops me from like going all in is trust. Like every so often you're going to see 'this got hacked', 'this was hacked.' $24 million lost, $36 million lost, $100 million lost and then we hope that we get it back and some people take the losses."

He notes that AI is making this problem cut both ways. Auditing tools are improving, which could eventually bring down the cost of security reviews from the $50,000 to $100,000 range that currently stops many independent projects before they ship. But bad actors are also using AI to scan old contracts for exploits at scale.

Voting Where the Emissions Go

The KAT token sits at the center of Katana's incentive model. When locked as vKAT, it gives holders voting power over where emissions flow across the chain. Right now that voting is limited to DEX liquidity pools, but the roadmap extends it to borrow-lend markets, yield tokenization positions on Spectra, and eventually Perps trading volume incentives.

Billy compares the mechanic to frequent flyer miles with route-voting privileges. A user who supplies liquidity to the ETH-USDC pool is going to vote for the ETH-USDC pool because that is where their money sits. The self-interest is the point. As long as a user keeps participating and voting, the emissions can follow their position indefinitely, rather than disappearing after a two-month campaign from the chain's treasury.

A product called avKAT will allow users to delegate their voting so the system auto-compounds without requiring bi-weekly manual input. Not every DeFi user wants to manage that cadence, and Katana is building options for both kinds of participants. There are also ongoing incentive campaigns with Binance and OKX running concurrently at the time of the episode, with large KAT token allocations on both platforms. Billy mentions he cannot access them personally due to regional restrictions, but describes them as a meaningful on-ramp for users in those regions who want to get early exposure to the voting system.

Saving Tokens with the Katana MCP Server

The MCP (Model Context Protocol) server Billy has been building for Katana is not live yet at the time of the recording, but he walks through how it works and why he built it. The core problem it addresses is token usage. Every time an MCP server starts up, it loads its full tool schema into the context window, which burns through API tokens quickly. For users on a standard Claude subscription, that overhead has been a real barrier.

Billy's solution is to run scripts in the background that pre-index data from Morpho markets and Merkle incentive feeds, then return only the top results rather than forcing the language model to do the scanning itself. A query about which Sushi pool has the highest incentives might use 200 or 300 tokens rather than thousands.

The tool is designed to serve two audiences. A newer DeFi user can ask which USDC vault has the best APY and get a ranked list of Morpho markets with current utilization rates. A more experienced user can ask for the best looping strategy and get back a recommendation that accounts for DEX fees, available liquidity, and how many loops are viable before slippage starts to hurt the unwind. The server can then help generate a smart contract to handle the rebalancing if the user wants to go further.

Billy says he built it because he uses Katana himself and found himself repeatedly running through Excel and Dune Analytics to make decisions he wanted to automate. The narrow application stack that makes Katana opinionated by design turns out to be exactly what makes the MCP server tractable. There is one DEX, one borrow-lend, one incentive distribution layer through Merkle. The surface area is small enough to scan without hammering an RPC node or burning a monthly token budget on a single prompt.