Plainly

Plainly is xCOR rebuilt for Robinhood Chain. The core loop is identical: a user writes a plain-English instruction, an intent engine turns it into a concrete on-chain action, and the agent executes it after the user confirms. What changes is the surface it acts on.

On Base, xCOR trades DeFi: swaps, lending, yield. Robinhood Chain (Arbitrum Orbit L2, USDG native) is built around tokenized real-world assets, primarily tokenized equities, alongside standard DeFi primitives. That gives Plainly a materially stronger pitch than a straight port: it becomes the plain-English interface to both tokenized stocks and DeFi in a single agent.

"Buy me $500 of tokenized AAPL." "Move my idle USDG into the highest-yield vault." "Sell half my tokenized TSLA and hold the proceeds in USDG."

One agent, one sentence, one chain.

The wedge is friction removal. Robinhood Chain is bringing a large, non-crypto-native audience toward tokenized assets. Those users don't want to learn swap routers, gas mechanics, or vault dashboards. Plainly is the layer that lets them do everything by typing a sentence, the same way they'd tell a broker.

• Say it. Plainly does it.
• Trade and earn on Robinhood Chain in plain English.
• The plain-English agent for tokenized assets and DeFi.

1. Input. User types an instruction in natural language.
2. Intent parsing. The instruction is resolved into a structured action: asset, direction, size, protocol/route, constraints.
3. Simulation. The agent builds the transaction(s), simulates against current state, and returns a plain-English preview with expected outcome, fees (USDG gas), and slippage where relevant.
4. Confirmation. User approves. Nothing executes without an explicit confirm step.
5. Execution. The agent signs and submits on Robinhood Chain and reports back in plain English.

The simulate-then-confirm gate is the product's trust mechanism and is treated as non-negotiable.

Tokenized equities. Buy, sell, and check positions on tokenized stocks native to Robinhood Chain, sized in USDG.

DeFi primitives. Swap between assets, deposit and withdraw from lending or yield vaults, check balances and positions.

Portfolio queries. Plain-English reads: "what am I holding," "how's my USDG earning," "what did that last trade cost me."

Everything settles in USDG, which simplifies sizing. Users think in dollars, the agent handles the rest.

Natural-language layer. Parses the instruction with a small, fast model (Claude Haiku 4.5). Output is a strict structured schema, not free text.

Intent / routing engine. Deterministic. Takes the structured intent and resolves it to concrete calls: which protocol, which route, what parameters.

Execution layer. Builds, simulates, and submits transactions on Robinhood Chain. Handles USDG gas, signing, and confirmation reporting.

Settlement. Robinhood Chain. Arbitrum Orbit L2, USDG-native gas.

The hard rule: the model proposes intent, deterministic code disposes. The LLM never has authority to move funds on its own.

• Frontend: React / TypeScript / Tailwind. Conversational UI with a live simulation preview.
• Backend: Supabase for auth, session, and transaction history.
• Intent model: Claude Haiku 4.5.
• Chain: Robinhood Chain RPC, USDG as base unit, Arbitrum Orbit-compatible wallets.

Phase 1. Intent engine. NL parsing → structured schema → deterministic routing.

Phase 2. Read + simulate. Balances, positions, prices, transaction simulation. Nothing executes yet.

Phase 3. Execution. Signing and submission behind the confirm gate, starting with a single tokenized-equity buy in USDG.

Phase 4. Frontend + landing. Conversational app UI, transaction history, and a premium landing page.