Each KPI is defined once and answered from the company’s own data. The approved numbers are checked against the source every week, so a wrong one shows up before it reaches a board deck. And when teams ask AI, they get that same number.
ARR, churn and margin mean something slightly different at every company in the portfolio.
The monthly pack is rebuilt by hand, from the app, the CRM and billing, and that’s where numbers drift.
Teams are already asking AI for the numbers. Nobody checks whether the answers are right.
What ARR, churn or margin means is written down with the people who own the numbers. AI drafts the definitions from the company’s data; your team approves them.
Every question is checked against the approved definitions and tables before it runs. The AI doesn’t guess what a column means.
The questions you approve are asked again every week and scored against the company’s own data. If a system or model change breaks a number, the Accuracy report flags it before anyone reports it.
Accuracy report
Week 45 · one portfolio company
23 of 24 approved answers match
Net revenue retention for October changed after last week’s billing update. Flagged for review before the monthly pack goes out.
Checks numbers, names and lists for closed periods, like last month or last quarter. Example data.
Every portfolio company gets its own workspace, connected to its own systems, with its own rules for who sees what. Sales sees sales. Finance sees finance.
Who asked what, which data the answer used, and what came back. For every person and every automated report.
Works with whichever AI the teams use today, and whatever they switch to next. The definitions and permissions stay put when the AI changes.
We set it up with you, one portfolio company at a time. Each company’s team runs it from there.
Step 1
Start where the monthly pack hurts most. Bring the five numbers you report every month.
Step 2
We connect its systems and write down what each KPI means, signed off by the people who own the numbers. Everything lives in the company’s own accounts.
Step 3
The team asks AI for those numbers, and the first Accuracy report shows which ones match.
Step 4
Same playbook, same KPI list as a starting point, each company on its own data.
The same layer runs each company’s own operations: the questions the team asks every day, the Monday report that lands before anyone asks, a heads-up when a number moves, and AI agents that handle the routine work. The numbers the parent sees are a by-product. Here’s what changes by role.
GM / COO
The numbers your parent company asks for come straight from your systems, defined the same way every month and checked every week. Nobody rebuilds the spreadsheet.
CTO
Access per team and per table, read and write kept separate, and every question on record. It works with whichever AI your people already use.
CMO
Ask which channel drove pipeline and get the same answer finance would give. Plus a Monday report on what moved, and a heads-up when a number you pick changes.
It started with
Business questions and automated reports on their data.
Then
They connected it to their AI through FrontierOps.
Now
People across the group ask their own questions, every day.
30
people using it, from a single data project
