Atomic roadmap - four directions in parallel

Atomic has been shipping. New design, a faster app, referral payouts denominated in dollars, an order book that reads the chain directly, routing v2 with position-open latency down 78% on average. The contracts haven't changed - that's the floor we don't move. What sits above the floor is the interesting question.

Short version: four directions, in parallel, on purpose.

Why four, not one

The simpler story would be picking one direction and going deep. We are not telling that story, because each of the four solves a different problem, hedges a different cycle, and reaches a different customer.

TradeFun removes us as the bottleneck on what gets listed in the first place. Algo traders need APIs and don't care about UI. Oracle perps unlock assets that an L2 spot DEX cannot physically list with enough depth. Prediction markets live off the news cycle and keep volume flowing when crypto sleeps.

Picking one of these is a fragility bet on a single customer and a single market mood. Building all four is what a protocol that wants to outlive cycles looks like.

This post is the map. The four sections below are roughly in shipping order - what's closest to launch first.

1. TradeFun - markets created the way you create accounts

PumpFun settled one thing: creating an asset should be as friction-free as creating a social account. No DAO committees, no three-month audits, no "come back next governance cycle." The protocol provides the rails. The catalog gets decided by the catalog's users.

We are applying that idea to leveraged markets. A user fills in a few fields, a pre-launch sales pool opens, and if the cap fills, two pools spawn: a trading pool and an isolated lending pool that backs leverage on the new asset. Boring markets die. Interesting markets accumulate volume. Creators and the platform earn a share of fees.

This removes us as the structural bottleneck. Atomic stops being the choke point that says "bring us $50M of liquidity and we'll list the pair." Atomic becomes the rails. Inventory shows up on its own.

MVP ships this week. Full breakdown alongside the launch.

2. API + Webhooks - Pine Script meets on-chain settlement

On a CEX, an algo trader writes a Pine Script, points it at the exchange's REST API, sets an alert, and the alert fires the trade. Table stakes there. On-chain it is nearly absent - and the few DEXes that have it route through middlemen who hold custody, which kills the only reason someone came to a DEX.

We are building REST API + Webhooks with two custody models: run the signer yourself, or fund a dedicated address whose key never leaves your hands. The platform never holds it. Pine Script alert fires a webhook, the webhook signs a transaction, Atomic settles.

Algo traders bring an order of magnitude more volume than retail and do not churn after a drawdown. Different customer, different revenue floor.

Full breakdown in two weeks.

3. Oracle-priced perps - answering the depth question

L2 liquidity is a wall. No router on Arbitrum, however well-built, gives you the depth of Binance. That isn't 0x's fault, or 1inch's, or Kyber's - the spot orderbooks just aren't there yet. You can complain about it, or you can build perps.

Perps with oracle pricing change the question. Price comes from the oracle, not the book - which means any asset can have leverage attached, including the long tail an L2 spot DEX cannot list. No slippage at the size most users actually trade. Funding rate as a new revenue stream.

This is the honest path for an L2 DEX to compete with a CEX on volume. Not "we'll match their depth one day" but "we'll ship the product they have, on our terms."

Full breakdown when the design firms up.

4. Prediction markets - a second product on a different rhythm

Polymarket settled the category question last year. Elections, sports, macro, "will GTA VI ship before Christmas" - the off-chain bookmaking industry is large enough to be a parallel financial system. The on-chain version is in its first decade.

We are integrating prediction markets as a module inside Atomic. Same wallet, same referral system, same infrastructure - resolved by oracles, settled on-chain. The point is not a Polymarket clone. The point is diversification: perps live on crypto volatility, lending lives on demand for leverage, prediction markets live on news. Different revenue sources, different clocks.

A deep-dive comes when we have specifics worth sharing.

What's already on disk

Some of this is built or partially built. The on-chain order book that landed in April is the depth view perps will use. Routing v2 is the latency floor TradeFun's launch flow needs to feel snappy. The isolated lending logic is what lets a TradeFun launch ship without breaking USDC.e LPs that back blue-chip pairs.

We don't believe in roadmaps that hand you a date. We believe in shipping cadence. Each direction shows up in the weekly ship log as pieces land. The TradeFun MVP is days away, the deep-dive runs with the launch. API and Webhooks follows.