Equity Engine

Ask about the data. Get a benchmark-backed answer.

Answers plain-English questions about workplace-fairness data with a benchmark table, a chart and a cited source, and shows its steps.

Built · pausedNo public access

Built for a workplace-equity benchmarking firm. Status verified 10 Oct 2026.

Book a call View all products

Read the transcript

01 The Wall of Workbooks

0:02 Before a client review, a consultant faces forty-tab workbooks, two benchmark files, and codes nobody remembers.

On screen: Illustration, not a product screen. Workbooks and benchmark files.

02 Point It at the Client

0:10 In Equity Engine, the consultant opens a folder for one client.

0:14 It builds a private knowledge base from that client's files and the benchmarks.

03 Ask Like a Human

0:19 Then the consultant asks in plain English, the way they'd ask a senior analyst, with no query language.

04 Watch It Work

0:26 First it reads the data dictionary, to learn what the codes mean.

0:30 Then it runs real analysis on the files, and shows its work.

On screen: Reasoning and the code it ran. It shows its work.

05 From Answer to Brief

0:34 The answer gives a verdict, a benchmark table, what to fix first, and a cited source.

0:40 A chart shows the gap: sponsorship trails Industry Best by 23 points.

0:46 One more request turns the answer into a one-page brief for the board.

0:51 It's written in the chat, ready to copy or export as Markdown.

0:55 Equity Engine, built by Mambo for a workplace-equity benchmarking firm.

The problem

Forty-Tab Workbooks. One Question at a Time.

Before a client review, a consultant digs through multi-sheet workbooks full of codes, lines them up against two benchmark files, then hunts for outside evidence.

Every follow-up question starts the spreadsheet work again.

What it does

Ask in Plain English. See the Work.

A consultant opens a folder for one client company, and Equity Engine builds a private knowledge base from that company's files and the benchmarks. Then they ask a question in plain English. It reads the data dictionary, runs real analysis on the files, checks a whitelist of research sources, and answers with a verdict, a comparison table, a chart and citations. One more request turns the answer into a brief to export or share.

01
A Private Knowledge Base per Client
Each folder holds one company's files, an industry's benchmarks, or the methodology alone.
02
Analysis It Shows
A reasoning summary and the code it ran sit behind a disclosure on every answer.
03
Benchmarks Built In
Answers compare against the Top 10 and the industry's best performer.
04
Cited Outside Evidence
Web search limited to a whitelist of research and government sources.
05
From Answer to Brief
Turn any answer into a structured brief, then export it or share the chat read-only.

What it doesn't claim

Answers, Not Decks. Offline Today.

It is not running today; the demo shows the version verified in March 2026.
Briefs are written in the chat and exported as Markdown; there is no separate report generator.
It works on the benchmark files loaded for each client, not on live HR systems.

About this demo

Recreated in Code, With a Fictional Retailer.

The demo rebuilds the real chat workspace with an invented home-improvement retailer and invented benchmark scores. Building the knowledge base and running the analysis are shortened and marked TIME-LAPSE. The benchmarking firm's brand is replaced with a fictional one.

In the demo

Five Chapters. Jump to Any One.

01 Play from 0:02: The Wall of Workbooks
The Wall of Workbooks
Dozens of tabs, two benchmark files and codes nobody remembers.
02 Play from 0:10: Point It at the Client
Point It at the Client
One folder, one private knowledge base.
03 Play from 0:19: Ask Like a Human
Ask Like a Human
A plain-English question, no query language.
04 Play from 0:26: Watch It Work
Watch It Work
It reads the data dictionary, then runs the analysis.
05 Play from 0:34: From Answer to Brief
From Answer to Brief
A verdict, a table, a chart and a source, then a brief to export.