AI Features

AI Across the Platform

Artificial intelligence is woven into every layer of Thrive, from signal generation to natural language queries to automated market analysis summaries. Rather than being a standalone feature, AI acts as an acceleration layer that makes every existing feature faster and more accessible. This page covers all AI-powered capabilities and how to use them effectively.

Thrive uses a combination of proprietary machine learning models for signal generation and large language models (OpenAI and Anthropic) for natural language interactions. The model used for each request is selected dynamically based on task complexity: simple queries use a fast model for instant responses, while deep analysis tasks use a more capable model for comprehensive output.

AI-Powered Insights

AI insights are contextual analysis summaries available on dashboard pages, asset detail pages, and within the Workbench. They provide plain-language interpretation of the data you are looking at, saving you the cognitive work of synthesizing multiple metrics into a coherent view.

Quick Insights

Quick insights are short, 2-3 sentence summaries generated on demand. They appear as expandable panels on asset pages and dashboard widgets. Click the Insight button on any data panel to generate a quick AI summary of the current state.

For example, clicking Insight on a funding rate chart might produce: "BTC funding rate is currently at 0.028%, which is 1.8 standard deviations above the 90-day mean. While not yet at extreme levels, funding has been trending upward for 3 consecutive days, suggesting increasing long positioning. Open interest has risen 8% over the same period, confirming fresh capital is entering on the long side."

Deep Analysis

Deep analysis is a comprehensive AI research report that examines an asset or market condition from every available data angle. It synthesizes derivatives positioning, on-chain flows, social sentiment, historical patterns, and active signals into a multi-section narrative with supporting data visualizations.

Deep analysis reports typically take 15-30 seconds to generate and produce 800-1,200 words of analysis with embedded data tables and chart references. They are cached for 1 hour, so regenerating the same analysis within that window is free.

AI Credit Costs

AI features consume credits proportional to the computational resources they require. Costs are transparent and displayed before execution so you always know the cost before confirming.

Market Analysis Summaries

The Dashboard home page features an AI-generated market summary that provides a high-level view of current market conditions. This summary is updated periodically and covers macro sentiment, top movers, notable funding rate developments, active divergences, and smart money trends. It is designed to be your first read of the day, giving you context before you dive into specific assets.

Market summaries are generated using a specialized prompt that aggregates data across all tracked assets and applies editorial judgment to surface only the most significant developments. The summary avoids speculation and focuses on observable data points with historical context.

Trade Idea Generation

The AI trade idea generator scans current market conditions and produces a ranked list of potential trade setups with supporting rationale. Each idea includes the asset, direction, suggested entry zone, stop loss level, take profit targets, and the data sources that support the thesis.

Using Trade Ideas Effectively

Alert Explanations

When a composable alert triggers, Thrive can generate a plain-language explanation of why it fired and what the current market context looks like. This is especially useful for complex multi-condition alerts where understanding which conditions converged, and what that convergence means, requires synthesizing several data points.

Alert explanations are opt-in and cost 5 credits per trigger. Enable them in the alert configuration under Notification Options → Include AI Explanation. The explanation is included in both in-app and email notifications.

AI in the Workbench

Beyond AI cells in notebooks, the Workbench integrates AI in several other ways that streamline the analysis workflow.

Model Transparency

Thrive is committed to transparency about its AI systems. Every AI output includes a disclosure of which model was used, what data was provided as context, and the confidence level of the response. You can click the model attribution tag on any AI output to see the full prompt and data context that was sent to the model.

Signal generation models are proprietary and trained on Thrive's historical data. Their architecture and training methodology are documented in the Trading Signals section. Natural language features use commercial LLM APIs with no user data retained for model training.

Next Steps