ConnectWiz Forms & Tables

Data that arrives structured — from your site, your chat, or anyone’s system.

A form builder that renders anywhere (including inside a live conversation), Airtable-style tables your AI and flows can read and write, custom fields on everything, and intake rules that act the moment data lands. One field engine underneath it all.

Forms on four surfaces Consent recorded with evidence Merge-safe deduplication
Form → table → CRM, live
Service request
Consent, with evidence
Requests table
Row written
Intake rule: tagged "warranty" · stage set
4
Surfaces a form renders on — hosted link, iframe embed, your own HTML, and inside a live chat
5
Ways rows enter a table — typed, CSV with preview, AI-extracted from text, web-crawled, or via the ingest API
11
Field types on one engine — powering contact fields, table columns and form fields alike
52
Languages the public form surface speaks — with internal settings never leaking to visitors

Form builder

One form, four stages — and a paranoid back office

Abstract 3D illustration: a glass form panel with a glowing consent checkbox, its data flowing into a structured table with a thread linking one column to a contact card

Hosted, embedded, or yours

Share the hosted link, drop the self-sizing iframe on your site, or keep your own HTML form and post it to our endpoint with a copyable snippet — your designers keep their pixels, we keep the plumbing.

Forms inside the chat

A flow sends the form into a live conversation; the answers come back as structured, branchable data — and you can build, preview and test-submit without leaving the flow canvas.

Spam, engineered away

An unreachable honeypot (screen-reader-safe, tab-safe) plus optional Cloudflare Turnstile — and when no CAPTCHA key is configured, the builder warns instead of silently shipping an open door.

Destination mapping, out loud

Each field maps to a table column or a contact field with one three-state rule — and the screen tells you when a field’s data is going nowhere, instead of letting you find out in a month.

Dedupe that refuses to guess

Email first, then phone with digit-normalized matching — and an ambiguous match creates a new contact rather than merging two strangers who share an office landline. Wrong merges are forever; we don’t make them.

Consent with receipts

The consent checkbox records the exact text, version hash, language, IP and context — and is deliberately not prefillable, because a link that ticks its own consent box proves nothing. UTM parameters are captured; sensitive fields aren’t.

Tables

Think Airtable — living next to your conversations

Price lists, service requests, inventory notes, event registrations — the operational data that usually lives in a spreadsheet nobody can automate. Here it lives where your flows and your AI can use it.

Abstract 3D illustration: a translucent data grid with kanban chips, an AI orb extracting rows from a paper sheet and a crawler thread pulling a row from a globe

Import with a preview

CSV import shows you what it understood before it writes — column matching, row preview, no surprise 400-row mistakes.

AI extraction & web-crawl fill

Paste messy text and AI extracts structured rows; point the crawler at a page and it pulls the data in. The spreadsheet-cleanup afternoon becomes a paste.

Kanban & ordering

Status columns turn any table into a board; manual ordering is real, not alphabetical fate.

Relations, gracefully

A cell can point at a row in another table — labels resolve automatically from the target’s first column, and a deleted target reads as an empty cell, never an error screen.

A guarded ingest API

External systems write rows through a token-secured endpoint — unknown keys dropped, status fields stripped at the boundary, body-size caps measured before parsing, and dual rate limits per token and per IP.

A node in the flow palette

Flows search, write and update tables mid-conversation — "check the price list, answer the customer, log the request" without a single external integration.

Custom fields

One field engine, three homes

Fields on contacts

Eleven field types — text, number, date, boolean, select, email, phone, URL and more — extend the contact card with the attributes your business actually tracks.

The same fields as columns

Table columns are the same field definitions — one engine, one validation, one rendering. A new field type lands everywhere at once, without migrations.

And as form fields

Forms draw from the same dictionary, which is why destination mapping is a matter of pointing, not of format wrestling.

Intake rules

Act on arrival — the record kind of automation

Flows start conversations; intake rules act on records. When a form submission matches your conditions, the record gets tagged, staged or escalated — before a human ever opens it.

Conditions with eight operators

Match on any field — equals, contains, empty, not empty and friends — and every matching rule runs, so two teams' rules never race each other for priority.

Tag, stage, notify

The actions are deliberately few and sharp: apply a tag, set a pipeline stage, send an email notification. Enough to route reality; too little to build a Rube Goldberg machine.

Fail-closed, twice

Both the engine and the saving gate refuse configurations they can’t evaluate safely — a typo can’t become a rule that silently does nothing for a month.

The honest fine print

What’s deliberately not here (yet)

No multi-step forms

Wizard-style forms and conditional field logic aren’t built. Long journeys belong in a flow, where each question is a step with real branching — that’s the honest recommendation.

No formula columns

Computed cells and cross-row rollups are a spreadsheet’s job. Tables here are operational storage for conversations and automation — scoped on purpose.

No public table views

Tables are internal. Publishing data to visitors happens through forms, the widget’s mini-apps or the catalog — surfaces built for the public, with public-grade guards.

Field types still growing

File-upload, multi-select and currency field types aren’t shipped yet — the relation type came first because linked data unlocked the most. We list what exists, not what demos well.

Forms & Tables FAQ

Before you build

More in the full FAQ, or ask us directly.

Four places: as a hosted standalone link, as a self-sizing iframe embed on your site, as your own HTML form posting to our endpoint (with a copyable snippet), and inside a live conversation — sent by a flow, answered in the chat, with the answers branching the automation. All four are fully localized in 52 languages.

Five ways: typed by hand, imported from CSV with a preview step, extracted by AI from pasted text, pulled from a web page by the crawler, or written by external systems through a token-secured ingest API with size caps and rate limits. Rows can relate to rows in other tables, and a flow can search, write and update tables mid-conversation.

Through an explicit destination-mapping screen: each form field maps to a table column or a contact custom field, with one three-state rule — unmapped fields flow to same-named targets, deliberately-emptied targets drop the data, named targets receive it — and the screen says out loud when a field is going nowhere. Duplicate matching is merge-safe: an ambiguous phone match creates a new contact rather than merging two strangers.

Small when-then rules that act on records as they arrive: when a form submission matches conditions, add a tag, set a pipeline stage, or notify by email. Every matching rule runs, and the engine is fail-closed — a rule that can’t be evaluated safely refuses to save rather than silently doing nothing.

Stop parsing inboxes. Start receiving structure.

Build a form, point its fields at a table, and let the intake rules do the sorting.