traig — a trading AI built on continuous strategy
A crypto trading AI that replaces the step-order grid bot with a smooth curve. Live on real capital for 9+ months, 100% non-custodial, with every trade and AI prompt published openly.
The first brand by the universal
traig is the first sub-brand built by the universal. the universal works at one end with high technology, and at the other end with everyday life that has nothing to do with technology at all. traig stands at the high-technology end.
A non-custodial trading AI running on real capital for 9+ months — a place where efficiency is pushed to its limit. But what the universal is really trying to fill in traig is not speed. It is the human seat. Opening every prompt, publishing every trade, leaving positions untouched until approved: on top of the highest technology, a margin for human verification and intervention is kept deliberately in place.
00 — Beginning
For a long time, the standard in crypto trading automation was the Grid bot. Divide a price range into equal intervals, place buy and sell orders at each slot in advance. Simple and intuitive, but when the market moves outside that grid the bot stays put. Inside the range it repeats orders of the same size at the same spacing. It does not read the grain of the market.
We started by hitting the exchange API directly, asking: can we draw a more natural curve with the same capital?
01 — The core idea
Recipe.
The strategy unit in traig. A piecewise linear function that maps price smoothly to allocation. The user defines a few control points; traig connects them with a natural slope.
Each segment has its own intensity. Steep where conviction is high, gentle where caution applies. Each segment can be set to buy-only, sell-only, or both. When price moves outside one recipe’s range, traig switches automatically to the next one. No need to manually stop and restart the bot.
A grid bot is stairs.
traig is a curve.
02 — Interface
Talk to your strategy.
The most tedious part of building a strategy was always the parameter form. Sliders, number boxes, spreadsheets. traig connects them with natural language.
Assistant Recipe updated. The allocation slope for the segment below $55,000 has been made 2× steeper. Compare the before and after curves before applying.
The AI translates intent into precise recipe edits. Before applying, the before and after curves are shown side by side. You approve only when satisfied. Until then, actual positions stay as they are.
03 — In the open
The AI layer is non-custodial too.
What sets traig apart is that the AI layer itself is under the user’s control. System prompts are fully public. Anyone can read, edit, and fork them. Internal tools are published as MCP specs, and users can bring their own model and API key.
System Prompt (excerpt) You are a trading strategy assistant. The user’s recipe has 3 segments: $40K–$55K, slope 0.8 (buy-only) $55K–$75K, slope 0.4 (both) $75K–$100K, slope 0.2 (sell-only) Max allocation: 85%. Return a structured JSON diff.
// This entire prompt is editable.// Fork it and build your own agent.If you cannot verify it,
you cannot trust it.
04 — Getting started
Automated trading in 3 steps.
01 / Connect
Exchange API key.
Your capital stays in your exchange account the entire time. traig never holds your assets.
02 / Configure
Build a recipe.
Design it through conversation with the AI, or start from a template and adjust. Backtest on historical minute data before going live.
03 / Automate
Maker-only execution.
Runs automatically around the clock. All orders are maker-only to reduce market impact and minimize fees.
05 — Track record
9 months, fully public.
Prove it with numbers, not words — that is the traig team’s principle.
Running 24 hours a day on real capital for 9+ months. Every trade is recorded, every decision is transparent. We publish the full record without cherry-picking any period. Strategy details, live dashboard, and signup are at traig.kr.
06 — Under the surface
It is complex inside.
24-hour operation without a single outage across 9 months runs on more than 10 microservices and zero-downtime deployment. That complexity lives where the user never needs to see it. What remains on screen is one curve and a few points.
07 — Who it’s for
Two kinds of users.
08 — With the universal
We use it first.
traig is a product the universal built to use internally from the start. We use the tools we make, before anyone else does. That studio principle holds here too. Nine months live on our own capital, and every friction that surface during operation goes back into the product.
traig is a different category from the other work the studio makes, but the way it is made is the same. Use it yourself. Refine it small. Record it honestly.
Next
Details at traig.kr.
Strategy explanations, live performance, whitepaper, and signup are all at traig.kr.