Most trading platforms are sold as a monolith: take the suite, migrate your users into it, and hope the vendor's roadmap stays aligned with yours. Bicut is built on the opposite bet — a trading stack should be a set of modules that snap into whatever you already run, one data model with seven doors into it.
This article is the map. Seven modules, what each one actually does, and the numbers behind them — all measured on a live brokerage running this exact stack in production, not projected from a slide.
A stack is not a bundle of features. It is one data model with seven doors into it — and you choose which doors to open.
The map
| Module | What it is | Headline numbers |
|---|---|---|
| Terminal | Trader workspace in the browser | 1M candles at 120 fps, 10 chart types |
| Core | Trading backend + risk engine | 198 REST endpoints, 102 WS events, 6 trading modes |
| Payments | Local methods, ramps, crypto | PIX, QRIS, cards, MPC wallet |
| Rooms | Live trading rooms with a host | up to 3,000 concurrent participants |
| Strategy Lab | Strategy catalog + backtesting | 3,300+ verified strategies |
| Voice Agent | Voice AI inside the terminal | server-side guardrails, human confirm |
| Signals | News scored into chart signals | deduplicated, explainable, on-chart |
Terminal — the surface
The Terminal is a complete trader workspace in the browser, driven by Flux, our own third-generation chart engine. One million in-memory candles at 120 fps, ten chart types, and a builder that assembles modular widgets — watchlists, order books, DOM ladders, screeners — into a branded terminal without a development cycle. It embeds over your backend; your users and their data never leave your platform.
Core — the engine room
Core is the part nobody screenshots and everybody depends on: accounts, instruments, orders, balances and a server-authoritative risk engine. Margin, PnL and liquidation are computed in one place, so every client — web, mobile, API — displays the same truth, and a liquidation preview is calculated before an order ever leaves the client. Six trading modes run in production: CFDs, perpetuals, spot, swaps, mempool trading and prediction markets. Take it as your backend, or keep yours and borrow only the risk layer.
Payments — the top of the funnel
Retail growth lives in markets that global card rails ignore or overcharge. Payments covers local methods — PIX, QRIS, cards, local fiat accounts — plus a crypto gateway across four network families and a non-custodial MPC wallet, all under one status contract: created, pending, confirmed, the same lifecycle for every rail. Deposits, trading and withdrawals become one reconcilable cycle instead of three vendors.
Rooms — the audience
Copy trading moves positions; it does not build an audience. Rooms puts a live host inside the platform — webcam, voice, a shared chart everyone watches, named cursors, chat the host answers out loud — for up to 3,000 concurrent participants, with recording and transcription built in. IBs and influencers bring their people into your platform, and session time follows.
Strategy Lab — the catalog
Strategy Lab ships with 3,300+ verified strategies on day one, each with a reproducible backtest. Users filter the catalog, backtest on history right in the browser — same result everywhere, down to the hash — and launch in demo, paper or live mode without leaving the platform. Strategies stop being content and become a product surface.
Voice Agent — the copilot
The Voice Agent is a voice AI that lives inside the terminal. It sees news, technicals and the account; it answers out loud and drives the chart — switch the symbol, add an indicator, mark a level — by voice. The part that matters: it prepares trades, and a human confirms every one. Risk limits are enforced server-side, where no prompt can talk them away.
Signals — the radar
Signals turns the news firehose into something a chart can display: headlines deduplicated, scored for impact, tied to the instrument and the event time, and landed on the chart as explainable markers — not a vendor black box. It also watches the on-chain long tail, with risk labels and allow/deny policies so the questionable never reaches your users.
How the modules snap together
Every module speaks the same data model, which is what makes the mosaic work: Signals land on Terminal charts, the Voice Agent reads them and prepares an order, Core checks the margin before it leaves the client, Payments funded the account in the first place, and the whole session might be happening inside a Room with three thousand people watching. One machine.
But the same contract works in reverse: unplug any module and run it alone over your existing platform. Start with one door — most teams do — and open the next when the first one pays for itself.
