129 QUESTIONS 700 SIGNALS · SOURDOUGH BREAD
TEMPERATURE = MONTHLY VOLUME 0940K
PLATE Nº 001 · 2026-08-10

OBSERVATORY · CURIOSITYGRAPH

Every search is a question in disguise.

CuriosityGraph shows you every question your market is asking about any topic — ranked by real search demand, organized the way people actually think, and handed to your content team or your AI agent. Stop guessing what your customers want. Read it.

70clusters charted5topics1,991,890volume indexed
THE PROBLEM

You are building for a customer you cannot see.

01

You guess what to make

Your content calendar and your roadmap come from a brainstorm, not from what your audience actually searched for this month. You ship the thing you think matters and hope you were right.

02

Keyword tools bury the signal

AnswerThePublic and the rest hand you hundreds of ungrouped phrases. You cannot tell which are the same question worded ten ways, which have real demand behind them, or which actually change a decision.

03

You cannot see how they think

Real curiosity branches — how-to, does-it-work, is-it-worth-it, this-versus-that, what-goes-wrong. You need the map of how a customer reasons through your category. No one draws it for you.

The most honest focus group in history is running right now. It is called search — billions of people typing exactly what they want, unprompted, at the moment they want it. You just have not been able to read it.

WHAT YOU GET

From a firehose of keywords to a map of intent.

Point CuriosityGraph at a topic. It reads thousands of real searches, merges the ones that mean the same thing, ranks them by summed monthly volume, and labels how each is asked. Here is an actual result from the live corpus — sourdough bread:

QUESTIONSEARCHES / MOMERGEDSIGNAL
how to make sourdough bread starter940,30025 phrasingstop demand
is sourdough bread nutritious374,30027 phrasingstop demand
is sourdough bread healthy374,30027 phrasingstop demand
is sourdough gluten free bread105,0006 phrasingstop demand
how to store sourdough bread46,1008 phrasingstop demand
is sourdough bread fattening22,200top demand

The top line — how to make sourdough bread starter — is 940,300 searches a month, 25 different phrasings collapsed into one. The lines below it are the objections and follow-ups people also ask — the doubts that decide whether they buy your product or trust your guide. That is not a keyword list. It is your customer's mind, ranked.

Every question above is real, pulled live from the corpus. We never invent data — when we don't have it, we say so.

WHO IT IS FOR

For everyone who has to answer "what does our customer actually want?"

Content & SEO teams

Fill the calendar with demand-backed briefs. Build topic clusters that own a category. Every article maps to real questions with real volume — and comes with its H2/H3 skeleton and FAQ already outlined.

Product & marketing

See what your market wants to know before they buy, and the objections that stop them. Write docs, FAQs, and positioning that answer real questions — and quietly cut your support load.

AI builders & agencies

Question intelligence inside your pipeline. Call research_topic from Claude, Cursor, or n8n and get structured, ranked questions back — at scale, for every client, without a human in the loop.

Founders & operators

Instant command of a niche. Understand what your audience is asking and produce content like you have a research team behind you — because now you do.

HOW IT WORKSPLATES I–VIII

Real data, eight steps, nothing invented. Every question is backed by real search volume; semantically identical queries are merged by meaning, not string-match; each cluster is verified before it earns its place.

I

Seed

A topic expands into thousands of real queries via DataForSEO Labs.

II

Filter

Interrogatives and comparisons become questions; the remainder is the tag pool.

III

PAA

People-Also-Ask questions harvested from the live SERP, one level deep.

IV

Cluster

Queries embedded with bge-m3 and merged by semantic identity.

V

Verify

A model confirms each cluster and names its canonical question.

VI

Tag

Real keywords attach to clusters by cosine similarity, weighted by volume.

VII

Rank

Clusters ranked by summed real search volume; PAA presence breaks ties.

VIII

Assemble

Themes, trees, and article briefs — cached as a compounding question graph.

BUILT FOR AGENTS

Your workflow wants an endpoint, not a dashboard.

CuriosityGraph is agent-first. Call it over MCP from Claude, Cursor, or n8n; hit the REST API from any language; or run the CLI. Structured, ranked question intelligence flows straight into your pipeline — first agent call in under two minutes.

MCP

Connect your agent in under two minutes.

mcp.curiositygraph.com
tools: research_topic · expand_question
       get_tree · get_tags · generate_brief

REST

OpenAPI-first. Async jobs, polling, HMAC webhooks.

POST /v1/jobs { topic, depth }
GET  /v1/jobs/{id}
GET  /v1/jobs/{id}/result
Read the docs →

CLI

Prints your key, runs a sample job.

npx curiositygraph "solar panels"
IN YOUR AGENT

Research the customer, then build — from your editor.

This is the age of agentic code. Point Claude Code, Cursor, or an n8n prompt node at CuriosityGraph and it pulls real customer questions, then writes the content around them. Paste one of these and go.

Pillar article, ranked by demandCLAUDE CODE · CURSOR
Use the CuriosityGraph MCP to research "cast iron
skillet care". Take the 8 highest-volume questions,
write a guide that answers each as an H2 — highest
demand first — and add an FAQ from the questions
people also ask. Note the monthly volume per section.
→ a publish-ready draft, grounded in what people actually search
A month of content from one topicCLAUDE CODE
Call research_topic on "sourdough bread" via
CuriosityGraph. Cluster the questions into 10 article
ideas, rank them by summed search volume, and lay out
a 4-week calendar — one pillar plus supporting posts —
with the target question and demand for each.
→ a demand-ranked content calendar, not a brainstorm
Answer the objections (FAQ component)CURSOR · REACT
Research "electric vs gas water heater" with
CuriosityGraph and pull the comparison and objection
questions. Generate a React FAQ component that answers
the top 12 — each with a short, accurate answer that
pre-empts the doubt.
→ an FAQ that closes the doubts before they cost a sale
Automate it, weeklyN8N · MCP
Every Monday: research_topic on our 5 product
categories, flag any question that newly crossed
1,000 searches/mo, generate_brief for each, and open
a draft in the CMS — no human in the loop.
→ fresh, demand-backed briefs every week, hands-free

Connect once — mcp.curiositygraph.com — or hit the REST API from any language. Every free account ships with a key.

THE ASSET

A corpus that compounds.

Every job you run enriches a shared graph of questions, volumes, and relations. Because results cache, popular topics cost pennies and answer in seconds — you are building an asset, not renting a dashboard.

70question clusters
5topics charted
1,991,890searches indexed
<2 minto first agent call
PRICING

Start free. Pay for what you chart.

Prepaid credits, no subscription to forget about. A quick job is 1 credit, standard 4, deep 10 — and warm topics you have already charted are nearly free.

Free

2 jobs

  • Two quick research jobs
  • Email-verified, no card
  • Full API + MCP access
Start →

Packs

$19–149

  • 20 / 60 / 220 credits
  • Valid 12 months
  • Quick 1 · standard 4 · deep 10
Buy credits →

Pro

$29/mo

  • 40 credits monthly, 2× rollover
  • Deep research mode
  • Scheduled re-research
Go Pro →

START HERE

Point at a topic. See your customer's mind.

Start free — 2 jobs, no cardRead the docs