Connects to the warehouse, turns a question in plain language into SQL, runs the attribution, and returns a chart with the reasoning attached.
Handles messy joins, time-series forecasting and causal inference, and watches the core metrics for movement worth a human looking at.
Handles messy joins, time-series forecasting and causal inference, and watches the core metrics for movement worth a human looking at.
import { NexusAgent } from '@nexus-ai/sdk';
const analyst = new NexusAgent({ role: 'data_analyst', model: 'nexus-dense-70b' });
const report = await analyst.execute({
dataSource: 'warehouse://prod-snowflake/sales',
query: 'Compare CAC and LTV across NA and EU for Q3'
});
console.log(report.dashboardUrl);
Published examples, reproduced verbatim. This page does not call a model.
Look at the last 90 days of paid acquisition across North America and Europe. Find what is driving the drop in return on ad spend, and tell me what to do about it.
Aggregated 1,420,000 campaign events across 6 channels. The drop is concentrated in one place: cost per click on Meta in Germany rose 34.2% over the period, against 4.1% elsewhere. Impression share fell from 41% to 26% over the same weeks, which is consistent with a competitor bidding into the same keyword set rather than with a change in your creative. Everything else held: conversion rate is flat within noise, and the US accounts are unchanged. Suggested move — shift the German budget to search placements on the channel where your impression share is still above 60%. On last quarter's conversion rates that recovers most of the lost efficiency; I would review it again in three weeks rather than treat it as settled.