Crosstab & tab book software for agencies & insights teams
TheAIneverdoesthemath.
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.
Build banners and filters from any variable — AND / OR
logic, labels that write themselves, significance re-tested — and
every table in the study follows.
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.
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.
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.
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
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
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