What is Cornilius — business analytics inside your AI agent

What is Cornilius

Consistency and follow-through, for your analytics.

Cornilius is business analytics inside your AI agent. It carries your metric definitions, playbooks, and open issues from one conversation to the next — so your agent works from the same rules and nothing slips.

Click a panel to continue. Arrow keys work too.

What is Cornilius, panels

The human problem with analytics

Analytics breaks in human ways. Each analyst does it a little differently. The same question can end in different conclusions — and some conclusions simply get lost. Cornilius takes these problems one by one:

Every analysis runs a little differently

PLAYBOOKS — one procedure for your agent to follow

The same metric, calculated differently across the business

METRIC DEFINITIONS — one definition, carried into every conversation

Issues slip away unresolved

DOMAIN OWNERS — every issue has an owner, so your agent can escalate to the right person

What Cornilius is

Cornilius solves this consistency problem with business analytics built into your agent. It holds your metric definitions, analytical playbooks, open issues, and prior findings — your agent reads from it and writes to it as it works. Each finding is stored as a structured issue, not a running log, so ask your agent a question about your business and it gets back what is relevant to that question, not every open issue at once. It sits behind your agent, keeps only the minimum — definitions, playbooks, issues, and findings — and has no direct access to your raw data sources.

YOU
YOUR AGENT
YOUR DATA SOURCES
CORNILIUS
DEFINITIONS
PLAYBOOKS
ISSUES
FINDINGS

Cornilius sits behind your agent — minimal data, no direct access to your raw sources.

Consistency: your definitions, applied

Encode a metric definition once and every conversation starts from it. Cornilius carries your data logic forward, so your agent applies the definition your team agreed on instead of reconstructing it from scratch.

WITHOUT

ROAS = revenue ÷ ad spend = 4.2

ROAS = new-customer revenue ÷ ad spend = 2.8

// two definitions coexist — unclear which one a report used

WITH

ROAS

new-customer revenue ÷ ad spend

OWNER: @maya_growth

applied in sessions 03, 07, 12 — the same definition, every analysis

Follow-through is memory

Analysis creates loose ends: an anomaly to watch, a number to re-check, a question for whoever owns the data. Cornilius records each one with a status, an owner, and the channel to raise it in. Your agent starts the next conversation knowing what is open and who it belongs to — so open items get raised, not lost.

refund spike — eu region

STATUS
WORSENING
OWNER
@jason_finance
CHANNEL
SLACK: #finance-alerts
ESCALATION
ESCALATED
LAST SEEN
MAR 18

checkout conversion dip

STATUS
REPEATING
OWNER
@maya_growth
CHANNEL
JIRA: GROWTH-142
ESCALATION
PENDING
LAST SEEN
MAR 11

data gap: mobile events

STATUS
STABLE
OWNER
@noa_data
CHANNEL
SLACK: #data-eng
ESCALATION
PENDING
LAST SEEN
MAR 04

Built to be yours

Cornilius holds instructions and findings, not your raw data — your warehouse remains the single source of truth.

PERSISTENT

Carries forward across sessions, models, and analysts.

ISOLATED

Your definitions and findings, not your raw data. Tenant-isolated by construction.

PORTABLE

Open MCP. Claude today, any MCP-capable agent next.

The point

One memory underneath your agent: definitions that hold, playbooks that repeat, open issues that come back until they are closed.

Self-serve setup. Connect Cornilius to your agent as an MCP server.

Get started

Connect Cornilius to your agent

Self-serve setup. Create your MCP credentials and add Cornilius to your agent as an MCP server.

Get started