AlphaSense

Overview

AlphaSense is a toolbox that Agents can use to tap into OpenGradient’s advanced and secure ML workflows in order to perform much more advanced actions and complex tasks, such as price forecasting, risk calculation or portfolio optimization. Thanks to our end-to-end AI network, agents can browse, use and schedule the execution of any open or closed-source AI models through our Model Hub.

Benefits

• Agents can tap into AlphaSense to perform much more difficult actions and complex tasks, like leveraging risk forecasting streams to shield portfolios from drawdowns and volatility.
• Applications on different chains can use AlphaSense to optimize their protocols or develop intelligent features.
• Developers building agents can tap into AlphaSense to empower their existing or new agents by leveraging our verifiable ML workflows
• Secure scheduling and execution of models is entirely handled by OpenGradient
• Model inference, input and output is secured through ZK and TEE proofs
• Pluggable into existing agents and frameworks like LangChain or OpenAI Swarm

Example use-cases

• Volatility prediction: Generates a constant feed of volatility forecasts for assets powered by ML models, can be leveraged by portfolio manager or trading agents to de-risk portfolio, AMMs to scale trading fees, or lending protocols to adjust their LTV ratio.
• Spot forecast: Creates a constant feed of spot returns forecast for assets powered by time-series models, can be leveraged by yield strategies to improve risk-adjusted returns.
• Sybil detection: Takes in a series of wallet addresses and outputs within a confidence interval which wallet addresses are likely to be Sybil accounts given past transaction history.
• Markowitz: AlphaSense that takes in portfolio holdings, volatility forecasts, and returns forecasts and generates optimal portfolio goal positions using mean-variance optimization.

Usage

AlphaSense is exposed as a set of tools that can be plugged into any agent. Advanced ML models can be plugged into any agent as a tool:

LangChain

import opengradient as og

spot_forecasting_model_id = 'QmY1RjD3s4XPbSeKi5TqMwbxegumenZ49t2q7TrK7Xdga4'

# Create spot forecasting tool for SUI/USD
spot_forecast_tool = og.mltools.create_og_model_tool(
    tool_type=og.mltools.ToolType.LANGCHAIN,
    model_cid=spot_forecasting_model_id,
    tool_name="SuiSpotForecast",
    input_getter=lambda: {"open_high_low_close": fetch_sui_price_history()},
    output_formatter=lambda out: f"The predicted price change is: {out['destandardized_prediction'][0] * 100}%",
    tool_description="Runs an ML model to forecast the price of SUI 30 minutes from now. Requires no input."
)

# agent can use spot forecasting tool as part of its execution
agent = create_react_agent(opengradient_llm, [spot_forecasting_tool])

Swarm

import opengradient as og

spot_forecasting_model_id = 'QmY1RjD3s4XPbSeKi5TqMwbxegumenZ49t2q7TrK7Xdga4'

# Create spot forecasting tool for SUI/USD
spot_forecast_tool = og.mltools.create_og_model_tool(
    tool_type=og.mltools.ToolType.SWARM,
    model_cid=spot_forecasting_model_id,
    tool_name="SuiSpotForecast",
    input_getter=lambda: {"open_high_low_close": fetch_sui_price_history()},
    output_formatter=lambda out: f"The predicted price change is: {out['destandardized_prediction'][0] * 100}%",
    tool_description="Runs an ML model to forecast the price of SUI 30 minutes from now. Requires no input."
)

After plugging the model into the agent, it can use it as part of its regular reasoning and execution flow, outsourcing complex tasks to specialized AI models.

Check our API Reference for further details.