Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.datagenie.ai/llms.txt

Use this file to discover all available pages before exploring further.

The home page covers why DataGenie exists. This page covers how it works — the lifecycle, the detection pipeline, the story anatomy, and the personas involved. Read this once and everything else in the docs will click into place.

The lifecycle

DataGenie is a closed loop, not a report. Data goes in once; stories come out continuously.

Onboard with GO

Point GO at a data sample and describe the business problem in plain English. GO scans the schema end-to-end, proposes KPIs and dimensions, picks the right time granularity per KPI, and produces a Blueprint — the complete operational definition of what the dataset monitors. No data engineering. Learn more →

Detect

From just 6 KPIs × 6 dimensions, DataGenie auto-creates and tracks 100,000s of metric combinations. A deterministic ensemble — Prophet, ARIMA, SARIMAX, LightGBM, XGBoost, plus a custom Fourier-based seasonality detector — runs continuously and picks the best forecast per KPI. No threshold tuning.

Connect

Related anomalies — across metrics and across data sources via the Nirvana algorithm — roll up into a single Top Story with root cause and contributors. You get one story, not a flood of alerts. Learn more →

Ask

Wisdom turns plain-English follow-ups into deterministic analytical pipelines. A master coordinator agent picks which of five services (Metric, Insights, Contribution Analysis, Forecasting, Scenario Planning) to run and in what order — and delivers the result as charts, tables, and natural-language summaries. Learn more →

Act

Route stories to Slack, Teams, Email, or JIRA on your cadence. Pin answers as QuickLook dashboards. Save recurring analyses as reusable Wisdom Skills. The insight finds the right person on the right channel.

Anatomy of a Top Story

Every autonomous story DataGenie produces follows the same five-part structure. Read one in 15 seconds; know exactly what to do next.
PartWhat it answersExample
WhenWhich period”Week ending 2026-04-18”
WhereWhich dimensional context”In US, on iOS”
WhatThe headline KPI change”Revenue dipped 11%“
What ElseRelated KPIs that debunk the obvious cause”Visitors +3%, bag rate +2% — so traffic isn’t the issue”
WhyThe Root KPI + 2–3 Contributors”Payment Completion Rate -21%, driven by PayPal on Mobile/iOS”
The structure means a director or VP reads a story once and knows where to act — no SQL, no dashboard-hunting, no analyst hand-off.

Autonomous, by design

No dashboards to build

Stories surface from data, not from the metrics you remembered to configure. The blind spot gap closes automatically.

No thresholds to tune

Anomaly detection is baseline-aware per KPI. Seasonality, zero-handling, and sparse series are handled automatically.

Connected, not isolated

Related anomalies — including across data sources — roll up into one story with root cause, not a flood of disconnected alerts.

Ranked by impact

Impact Score puts the story that actually moves your business at the top. Filter Presets let you scope detection to your team’s focus.

Deterministic core, scoped AI

The autonomous half of DataGenie is entirely deterministic — statistical and ML models running on your data, no generative AI in the loop. The agentic half (Wisdom) uses LLMs only for question interpretation and natural-language summarization, never to invent numbers.

Zero raw data to LLMs

Language models only see aggregated KPIs and dimension values. Row-level data never leaves your environment.

Deterministic numbers

Detection, attribution, forecasting, and scenario planning run on proven algorithms. Same inputs, same outputs — always.

Human-in-loop

Wisdom plans, runs deterministic services, and presents results for review. You validate; DataGenie doesn’t guess.

Guardrails you configure

Three surfaces per dataset encode your business rules so Wisdom and Top Stories answer within them — not around them.

Domain Knowledge

KPI definitions, business rules, dimension aliases, exclusions. Wisdom respects these on every answer.

Wisdom Skills

Packaged, reusable analytical pipelines in plain language. Invocable with arguments, deterministic output.

Business Events

A documented timeline of real-world context — promotions, outages, seasonality — so anomalies are explained with recorded causes, not invented ones.

Who uses what

DataGenie is built for business users — not developers. Different personas enter the platform from different doors.
PersonaPrimary entry pointSecondary
Business leader (VP, director, exec)Top Stories + QuickLooks on the home screenWisdom for ad-hoc follow-ups
Analyst / operatorWisdom + Explorer + Top StoriesFilter Presets, KPI Attribution
Data owner / adminDatasets + Anomaly DetectionGO for onboarding, RBAC
Workspace adminRBAC + SSO + Users & MappingsSecurity & Trust

The handoff map

DataGenie is a flow, not a collection of tools. Each feature hands off to the next.
Start hereArc stageHand off to
GOOnboard — turn raw data into autonomous monitoringTop Stories, Datasets
Top StoriesDiscover — autonomous anomaly narrativesWisdom, Explorer, Alerts
WisdomAnswer — agentic follow-ups, forecasts, scenariosDashboards, Wisdom Skills
ExplorerValidate — multi-dimensional verification of a WhyFilter Presets, Datasets
DashboardsShare — conversational canvas + QuickLooksRBAC
DatasetsConfigure — what gets monitored, how, and by whomAnomaly Detection, Nirvana
RBACGovern — three-tier access (Global → Tenant → Dataset)All features

Zero rip-and-replace

DataGenie sits on top of the warehouse you already have. Supported sources include Snowflake, BigQuery, Redshift, Databricks, PostgreSQL, and Azure SQL. No data movement, no new pipelines, no months of setup.

Connect

Point DataGenie at your warehouse. Read-only credentials; no data leaves your environment.

Configure with GO

GO scans, proposes, asks for domain context, and produces a Blueprint. Hours, not months.

Go live

Stories start surfacing immediately. Autonomous monitoring from day one.

From PoC to production in 8 weeks

The typical path from first connection to production rollout.

Weeks 1–2 · Setup

Agreements, provisioning, success criteria defined.

Weeks 3–4 · Deploy

Instance configured, first use case onboarded via GO, handoff docs delivered.

Weeks 5–6 · Validate

Active support, usage monitoring, feedback loops. Stories are landing.

Weeks 7–8 · Review

Assess against the criteria set in week 1. Document findings. Plan production scale-up.

Go deeper

Autonomous Insights

The detection engine at scale.

HyperConnected

Multi-source insight without ETL.

Responsible AI

Why DataGenie is safe for enterprise use.

Quickstart

Your first insight in minutes.

Core concepts

The vocabulary behind everything above.

Use cases

Concrete playbooks by business problem.