How Askflot Works
Askflot turns static questionnaires into adaptive, intelligent conversations. Learn about our 4-pillar conversational engine, 11 native validated types, webhooks, and REST APIs.
Jump to section
01. The 4 Pillars of the Engine
Multi-Entity Extraction
When a respondent shares a paragraph or voice note, Askflot simultaneously extracts multiple data points (e.g. role, budget, timeline, pain point) in a single turn instead of asking repetitive sequential questions.
Finite
LLMs tend to loop or ask infinite questions. Askflot enforces a strict question limit and terminates automatically as soon as required fields are collected or the budget limit is reached.
Deterministic Steering
The conversation is steered by business logic, not random hallucination. The engine mirrors the respondent's exact situation before formulating the next targeted question.
Discovered Metadata
When a respondent mentions valuable context that was not in your original schema (e.g. competitor name, urgent deadline), Askflot flags it as Discovered Data for 1-click schema expansion.
02. 11 Native Validated Types
Every field defined in a schema maps to a strict native data type. Values extracted by the AI are coerced and validated programmatically before database insertion:
| Type | Description | Validation & Normalization | Collection Mode |
|---|---|---|---|
| text | Short text strings (names, roles, titles) | Trimmed, whitespace stripped | natural |
| long_text | Narratives, detailed answers, stories | Preserves multiline formatting | conversational |
| number | Integer or float values (quantities, budgets, ages) | Regex validated, cast to int/float | natural |
| boolean | Binary yes/no, true/false decisions | Coerced from yes/no/oui/non/1/0 | natural |
| date | Calendar dates | ISO-8601 normalized | natural |
| Email addresses | Downcased, strict regex validation | natural | |
| phone | Phone numbers | Normalized format | natural |
| url | Website links | Valid URI scheme checked | natural |
| single_choice | One selection from options | Matches enum list | structured |
| multiple_choice | Multiple tags / options | Array normalization | structured |
| rating | Satisfaction score | Clamped automatically between 1 and 5 | structured |
03. Embedding on your Website
Embed your conversation seamlessly into any web page, blog, or modal using our responsive embed format:
<iframe src="https://askflot.com/c/YOUR_CONVERSATION_SLUG?embed=1" width="100%" height="650" frameborder="0"></iframe>
04. Webhooks & REST API
Whenever a respondent completes a conversation, Askflot dispatches an HTTP POST payload to your configured Webhook URL in real-time:
{
"event": "form_submission.created",
"form_id": 42,
"form_slug": "churn-audit",
"form_title": "Customer Cancellation Feedback",
"submitted_at": "2026-09-25T18:30:00Z",
"completion_time_seconds": 64,
"data": {
"cancellation_reason": "Switched to alternative for team collaboration features",
"competitor_chosen": "Linear",
"rating": 4
},
"ai_insights": {
"sentiment": "neutral",
"urgency": "medium",
"summary": "Customer liked the speed but required native team workspaces."
}
}