Skip to content
QUICK AI SEARCH
QUICK AI SEARCH

The intelligence engine, not another search box

A query becomes a run of independent reasoning loops: retrieval, claim extraction, cross-source verification, graph and timeline construction, synthesis. The result is an answer that shows its evidence, labels its confidence, surfaces contradictions instead of hiding them, and can keep itself updated for as long as you care about the question.

What do you want to discover?

Fast answer with citations
The search runs for real. Without a subscription you see the live intelligence activity and a preview; subscribers see everything. Plans
KnownSupportedInferredPredictedUnknown

Cited claims with confidence labels

Every statement is broken into atomic claims, each checked against independent sources and labeled Known, Supported, Inferred, Estimated, Predicted or Unknown.

contradictedvsverifiedtemporal supersession

Contradictions surfaced, not averaged

When sources disagree you see both claims, the severity, and whether time, scope or source quality explains the conflict.

Timelines and knowledge graphs

Dated events with stated precision, and entities and relationships you can explore as a graph with claims and sources as layers.

re-run every 6hnew claimconfidence shift

Living knowledge that keeps updating

Monitor a question and the system re-runs it on a schedule, records what changed, and tracks confidence over time.

How a run works

the full architecture
  1. 01 query
    Query

    Intent is classified and split into sub-questions; a budgeted plan picks the loops.

  2. 02 retrieval
    Retrieval

    Search providers, page fetches and the knowledge graph, each scored for reputation, freshness and independence.

  3. 03 claims
    Claims

    Evidence spans become atomic claims with provenance back to the exact fetch.

  4. 04 verification
    Verification

    Each claim is tested against other sources: supports, contradicts, neutral, with quotes.

  5. 05 synthesis
    Synthesis

    The answer is written from verified claims with citations, a calibrated confidence, and a list of what remains unknown.

Every loop is budgeted, timed and shown live in the activity rail while your search runs. Failed loops degrade the result; they never fabricate it.

Plans

No free tier. Pick a plan, approve the subscription on PayPal, and your account is created when the payment is confirmed. Cancel anytime.

Explorer

$18/month

Cited answers with confidence labels for everyday questions.

  • 200 searches per month
  • No Deep Research
  • No monitors
  • Modes: Quick, Explain, Verify, Discover
Choose Explorer
Recommended

Analyst

$38/month

Deep Research, comparisons, timelines, graphs and monitors.

  • 600 searches per month
  • 40 Deep Research runs per month
  • 5 living-knowledge monitors
  • All modes: Quick, Deep Research, Compare, Timeline, Knowledge Graph, Predictive Signals, Explain, Verify, Discover
Choose Analyst

Intelligence

$98/month

Everything, at volume, with the intelligent path on every mode.

  • 2,000 searches per month
  • 200 Deep Research runs per month
  • 25 living-knowledge monitors
  • All modes: Quick, Deep Research, Compare, Timeline, Knowledge Graph, Predictive Signals, Explain, Verify, Discover
  • Priority: the intelligent path on every mode
Choose Intelligence

Billed monthly through PayPal. Cancel anytime; access continues to the end of the paid period. No free tier.

Full comparison

Questions

Is there a free tier or trial?
No. Anyone can run a search from the front page and watch the intelligence activity, but the full result (answer, evidence, sources, graph, timeline) is shown only to subscribers. Anonymous visitors get up to three preview searches per day.
How does billing work?
Plans are billed monthly through a PayPal subscription. You approve the subscription on PayPal and your account is created only after PayPal confirms the first payment. You can cancel at any time from your account page or in PayPal; access continues to the end of the paid period.
What counts as a search?
Every run started from the search box counts as one search in the mode you chose, including follow-up questions and escalations. Deep Research runs count both as a search and against the separate Deep Research allowance. Scheduled monitor runs do not count against searches.
What is Deep Research?
Deep Research runs the full multi-loop path: iterative retrieval, claim extraction, cross-source verification, contradiction detection, graph and timeline construction, and a synthesized answer with calibrated confidence. It takes longer and reads far more sources than a Quick run.
Can the answer be wrong?
Yes. The system shows its evidence, labels every claim with an epistemic status, and lists what it could not establish so you can judge for yourself. Verify anything important against the cited sources before acting on it.
What happens when I hit a limit?
Searches, Deep Research runs and monitors are counted per calendar month. When an allowance is used up the run is refused with a clear message; you can upgrade from the account page and the new limits apply immediately after PayPal confirms.