question engine for ai agents

Your model has
blind spots.

Every LLM defaults to "How can I help you?" — a question that extracts zero signal. BetterAsk generates precision questions that fill your model's conceptual gaps. 6 archetypes. 607 battle-tested questions. One API call.

curl -X POST https://betterask.dev/api/generate \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "archetype": "philosopher",
    "context": "onboarding a new user",
    "depth": 8
  }'

// → 200 OK
{
  "question": "What's something you wish more tools understood about you without being told?",
  "archetype": "philosopher",
  "score": 8.7,
  "signal_density": 0.91
}
free tier — 100 calls/day — no credit card
// the problem

Agents ask bad questions.
Humans give empty answers.

Your AI asks "What are you looking for?" and gets "I don't know, something cool." That's not a user problem — it's a question problem.

Socrates figured this out 2,400 years ago: the quality of an answer is bounded by the quality of the question. BetterAsk encodes that insight into an API. Six archetypes of inquiry, each designed to extract a different dimension of signal from a human — so your model can build a richer internal representation of who it's talking to.

607
Battle-tested questions from END SMALL TALK
6
Question archetypes — each targets a different signal band
0.91
Avg signal density vs. 0.23 for generic prompts
📕
END SMALL TALK
by Cory Stout — the research behind the engine
// archetypes

Six Dimensions of Inquiry

Each archetype targets a different signal band. Combine them to map a human's full conceptual surface.

01
Philosopher
archetype: "philosopher"
Probes meaning, values, and worldview. Maps the belief layer.
"What belief did you hold five years ago that you've since abandoned?"
02
Time Traveler
archetype: "time_traveler"
Navigates past/present/future. Reveals trajectory and inflection points.
"What would 10-year-old you be most surprised about your life today?"
03
Mirror
archetype: "mirror"
Surfaces self-perception gaps. Extracts the delta between identity and behavior.
"What do your friends come to you for that you never advertise?"
04
Challenger
archetype: "challenger"
Tests assumptions. Finds where confidence exceeds evidence.
"What popular opinion do you think is completely wrong?"
05
Storyteller
archetype: "storyteller"
Elicits narrative. Narrative encodes more context per token than declarative statements.
"What moment completely changed the direction of your life?"
06
Alchemist
archetype: "alchemist"
Reframes and transforms. Converts failure/loss into usable signal.
"What failure are you now most grateful for?"
// pipeline

Three Steps to Signal

1
select archetype
Pick the signal band you need. Or let auto-select choose based on context.
2
generate question
Engine returns a scored question optimized for your context + depth target.
3
extract signal
Human answers. Your model ingests high-density context. Blind spots shrink.
POST
/api/generate
Generate a precision question
POST
/api/score
Score any question (6 dimensions)
GET
/api/archetypes
List all archetypes + metadata
POST
/api/register
Free API key — instant, no card
// pricing

Scale as You Learn

free
$0
100 calls/day
  • all 6 archetypes
  • generate + score
  • no credit card
get key
scale
$49/mo
50,000 calls/day
  • custom archetypes
  • webhook events
  • priority support
subscribe
volume
$199/mo
unlimited
  • dedicated endpoint
  • custom corpus
  • slack channel
subscribe