Crosstab & tab book software for agencies & insights teams

The AI never does the math.

A fielded survey becomes a live tab book. Every figure comes from the statistical engine — weighted, tested, cited to its table. Answers you can put your name on.

The AI narrates · the engine does the math · never the other way around
What it does ↓
Tabulation, the way researchers work

Drag a variable. Recut the book.

Build banners and filters from any variable — AND / OR logic, labels that write themselves, significance re-tested — and every table in the study follows.

Drag · drop · 291 tables recut · letters re-tested
See the book ↓
The AI does the work · the engine does the math

Ask for the deck.

Tell the AI what you need. It builds the banner points, cuts the tables and hands you a PowerPoint — every figure computed by the engine and cited to its table.

Builds versions · custom tables · native PowerPoint charts
See the AI layer ↓
MCP for your AI stack · early access

Your agents. Our numbers.

Point Claude, Claude Code or your own agent at FlexTabs over MCP. It gets weighted tables with significance letters — never raw respondents, never its own arithmetic.

10 tools · your own API keys · org-scoped · nothing respondent-level
See the exchange ↓
The FlexTabs Tables view: a weighted crosstab with significance letters
What it is

FlexTabs is crosstab software.

Your survey becomes a live tab book — cut, weighted, significance-tested — with charts, exports and an AI layer that never does the math. Everything here ships today unless marked early access.

Tabulate

  • Live book from your tab spec
  • Banners from any variable — drag & drop
  • AND / OR / AND NOT logic
  • Filters across the whole book
  • Swap or build any base
  • Nets, nested nets, custom tables
  • Hide, rename, reorder rows & columns

Test & weight

  • Z-tests on percents, T-tests on means
  • Overlap-corrected when columns share respondents
  • Effective base for weighted tests
  • Two confidence levels, letters in the cells
  • Low-base flags
  • Study, named or unweighted weights
  • Means, medians, std dev, NPS rows

Chart & export

  • Bar, stacked, pie, donut, line, dot
  • Significance markers carry into charts
  • Your client’s theme, not ours
  • Excel tab book — pick tables, banners, sig levels
  • Native PowerPoint & Excel charts
  • Several tables to one chart grid

Ask

  • Chat that cites the table and row
  • Builds banner points and custom tables on request
  • Hands back a PowerPoint
  • Stories: ranked findings, corrected for false discovery early access
  • MCP: 10 tools, your own scoped keys early access

Run it

  • Waves as versions — batch upload
  • Originals locked; duplicate to edit
  • Versions badged System · Human · AI
  • Organizations, roles, audit trail
  • Shareable workspace links
  • Runs on Microsoft Azure
The product

One workspace, from fieldwork to the room.

Crosstab tables with significance letters
The tab book

Your study arrives ready.

b³ programs and fields the survey; the tab spec becomes your live book the day fieldwork closes — every question, banner point and weight in place, significance already tested.

Drag any variable into a banner — AND / OR — and all 291 tables recut
Column tests at your study’s levels, letters in the cells, weighted bases
Watch the film →
A chart card: the biggest weeknight driver, drawn from Table 57
Charts & decks

Ready for the room.

Any table becomes a chart in one click, in your client’s theme, and lands in Excel or PowerPoint as a native chart — nothing retyped.

An Excel tab book with a picker for exactly which tables and banners
PowerPoint charts the room can still edit
See it in the film →
Insights chat: a question and its cited answer
AI layer Stories · early access

AI you can check.

Ask the study anything. The AI narrates, the engine does the math, and every answer carries the stat line and a deep link to its table. Stories reads the whole book first and ranks what’s worth your time.

Verbatim stat line and deep link on every answer
Stories: every cell scored, corrected for false discovery
What Stories is →
Infrastructure Early access

Your agents. Our numbers.

FlexTabs speaks MCP. Point Claude, Claude Code or your own agent at a workspace with a scoped key and it gets weighted, sig-tested tables — never respondent-level data, never its own arithmetic.

10 tools · your own API keys · organization-scoped
The same engine the app uses
How the keys work →
Why teams trust it

Built so you can defend every number.

Fast answers are easy. Answers you can put your name on are the product.

The AI never computes a number. Every numeral comes out of the engine
Significance the way you’d run it yourself. Z and T tests, overlap-corrected, effective base when weighted, two confidence levels
Every finding traces to its table. Deep-linked to the row · false discovery corrected, q ≤ .05
Checked against Wincross, cell for cell Runs on Microsoft Azure Organization-scoped access & roles Full audit trail Questions → FAQ
The film

FlexTabs in 50 seconds.

From fieldwork closing to the finished deck — the product film. Silent by design, so it plays clean on a phone across the table.

FLEXTABS · THE PRODUCT FILM

See it on a study you know.

We’ll walk your team through a live book — cut, tested, charted, and narrated — and answer the follow-ups in the room.

Book a demo
FlexTabs
Now taking demos
b³ Intelligence
Beyond the black box