Forecasting Built for
AI Agents
Give your AI agent accurate, production-ready time series forecasts via MCP. Connect Claude, ChatGPT, Cursor, or any MCP client in minutes. REST API also available.
Also available as a REST API
Everything You Need for Accurate Forecasts
Our API handles the complexity so you can focus on decisions, not data science.
AI Agent Native
MCPConnect Claude, ChatGPT, Cursor, or any MCP-compatible client. Your agent calls submit_forecast, polls status, and retrieves results - no custom integration code needed.
Configurable Horizons
Forecast from 7 to 364 days ahead. Choose the exact horizon your business needs.
Multiple Data Intervals
Support for 5-minute, 15-minute, 30-minute, hourly, daily, and weekly intervals.
Event-Based Forecasting
Include holidays, promotions, and custom events so your model understands demand drivers.
Location-Aware Modelling
Provide latitude and longitude to automatically incorporate local weather data into your forecasts.
Timezone Support
Full IANA timezone region support ensures your forecasts align with local business hours.
Exogenous Variables
Pass additional external signals like pricing, marketing spend, or foot traffic to improve accuracy.
Anomaly Handling
Handles anomalies automatically to maintain stable, accurate forecasts across changing conditions or irregular events.
Webhook Notifications
Get notified instantly when your forecast is ready. No polling required.
How It Works
Three steps from raw data to production forecasts.
Tell Your Agent to Forecast
Describe what you need in plain language. Your AI assistant - Claude, ChatGPT, Cursor, or any MCP client - handles the rest.
Agent Submits and Waits
The agent calls submit_forecast via MCP, polls for completion, and streams progress - all without you writing a line of code.
Results in Your Workflow
Forecasted values are returned directly into your agent conversation, ready to analyse, summarise, or act on immediately.
Submit Historical Data
POST your time series data to the /tasks endpoint with your desired configuration.
AI Processes Your Forecast
Our models analyse patterns, seasonality, and external factors. Receive a webhook when complete.
Retrieve Results
GET your forecast results with point estimates, ready for integration into your systems.
Connect in Two Lines of Config
A Few Lines of Python
Add AgentForecast to your MCP client and start forecasting immediately.
// Claude Desktop, ChatGPT, Cursor, or any MCP-compatible client
{
"mcpServers": {
"agentforecast": {
"type": "http",
"url": "https://www.agentforecast.ai/mcp",
"headers": {
"Authorization": "Bearer sk_your_api_key_here"
}
}
}
}"I've got 12 months of daily sales data in this spreadsheet. Can you forecast the next month?"
Calls submit_forecast with 365 data points → polls get_forecast_status → calls get_forecast_result
"Here are your 30-day demand forecasts: 1 May: 2,840, 2 May: 2,103, 3 May: 1,976 ..."
# pip install requests
import requests
API_URL = "https://www.agentforecast.ai/api"
API_KEY = "sk_your_api_key_here"
headers = {"X-API-Key": API_KEY}
# 1. Submit a forecast task
task = requests.post(f"{API_URL}/tasks", headers=headers, json={
"payload": {
"parameters": [{"parameter": "forecasting_length", "value": 43200}],
"data": [{
"dataset_type": "feature",
"values": [
{"datetime": "2025-01-01T00:00:00Z", "value": 142.5},
# ... more historical data
]
}]
}
}).json()
task_id = task["data"]["task_id"]
# 2. Check task status
status = requests.get(f"{API_URL}/tasks/{task_id}", headers=headers).json()
# 3. Retrieve results
result = requests.get(f"{API_URL}/tasks/{task_id}/result", headers=headers).json()
for point in result["data"]:
print(f"{point['datetime']}: {point['value']:.1f}")Ready to Start Forecasting?
Create an account and make your first forecast in minutes.