From Idea to Execution
How a prompt becomes an app, a deployment, and live portfolio state in OpenPond.
From Idea to Execution
This page explains the main strategy-development path in OpenPond.
The short version is simple: start in /agents/chat, turn the idea into an app, deploy it from the workspace, and then review the live result in portfolio surfaces and /trade.
1) Start in /agents/chat
Open Agents Chat and describe the strategy or tool you want. The best prompts are concrete about the market, signal, action, and risk controls.
Examples:
I want a BTC bot that reacts to major headlines but only buys when RSI is oversold.
Create a BTC strategy that trades escalation headlines and exits on ceasefire confirmation.
2) Turn the idea into an app
In OpenPond, the app is the main strategy container. It is where code, AI history, deployments, logs, and strategy activity come together.
For internal apps, OpenPond stores the code in an OpenPond-managed repo so the strategy has real file history and can keep evolving. For GitHub apps, OpenPond deploys from your repo instead.
3) Deploy from the workspace
Use the workspace, especially the right sidebar and deployment surfaces, to ship the current version of the app. A deployment is the runnable checkpoint for the strategy.
OpenPond primarily uses preview for iteration and verification, and production for promoted live versions.
See also: Deployments & Tools
4) Run tools, schedules, and backtests
Once the app exists, the next loop is straightforward: create or update the app, run tools or backtests, review the result, change the code or prompt, and deploy again.
Once deployed, the app can expose tools over HTTP, run on schedules, react to external market or event inputs, and use OpenPond-managed wallet and policy controls.
5) Review live portfolio state
Strategies should emit structured activity while they run. With OpenTool, they can use store() to persist decisions, execution attempts, lifecycle changes, and status updates.
OpenPond then derives additional state from strategy activity and venue data, including tradeFacts, positions, snapshots, and PnL context. That is what lets you review the result later in your portfolio surfaces and trading views.
See also: OpenTool Store Events
6) Keep one continuous loop
The point of this architecture is to avoid disconnected systems. The same app can move through the full cycle from prompt to code, deployment, live activity, portfolio state, and iteration.