What is Desearch — forecasting research data?

Web Search returns titles, snippets and links; Extract provides cleaned page text. AI Search adds synthesized output with sources, while X Search brings posts and conversations as another evidence stream. Combine the APIs while preserving source selection and timestamps.

The Numinous example describes retrieval during isolated agent evaluation, not an API-generated calibrated probability or a trade. Keep retrieval, interpretation and forecast output distinct, and budget each returned item, search, page or post under its own unit. Larger workloads can negotiate enterprise limits and service terms.

Who it is for

Developers adding current evidence to forecasting agents.
Researchers retrieving documents for market theses.
Builders combining web and social search.

Features and coverage

Web Search
Retrieve titles, snippets and source links.
AI Search
Request synthesized answers with cited sources.
Page extraction
Retrieve cleaned source text.
X Search
Query posts and conversations separately.
Composable retrieval
Connect search and extraction in an evidence workflow.
Numinous case study
Inspect evidence retrieval in isolated forecasting-agent evaluation.
Usage units
Each API has its own billing unit.
Enterprise capacity
Negotiate higher limits and service terms for larger workloads.

Workflows with Desearch — forecasting research data

Research an event

Search the exact question, fetch primary documents and retain their dates with the forecast.

Separate research stages

Keep retrieval, interpretation and probability output auditable.

Budget by unit

Estimate searches, returned items, extracted pages and posts separately.

Pricing and total cost

Reference rates per 1,000 units are AI Search US$0.40/items, Web Search US$1/searches, Extract US$0.50/pages and X Search US$0.15/posts. Console shows account-specific rates.

Fees and usage costs

X Search has a 10-post minimum per request, capped by the requested post count.

Getting started

Before you start: Retrieved evidence still needs interpretation and probability modeling.

Create an API account and choose evidence-collection endpoints.
Run a small request, inspect sources and review usage in Console.
Integrate retrieval with your own forecasting logic.

The service supplies evidence, not a standalone probability or trading bot.

Read the Numinous integration case study

Useful links

Review search and extraction rates

What to check

Retrieved evidence still needs interpretation and probability modeling.
The old inventory name does not describe the current case-study destination; Numinous scoring is separate.
Select reliable sources and preserve timestamps.

A closer look

Ratings & reviews

Editorial rating

3.5/ 5

0 user reviews

PredictionToolkit editorial

Our take on Desearch — forecasting research data

Search and extraction APIs with distinct billing units are useful forecasting components. Evidence is not a calibrated prediction, and request minimums affect actual costs.

Based on documented features, setup and access terms. How we rate

User reviews 0

The main score switches to user ratings after 5 published reviews.

Be the first to share your experience

A specific example helps the next person choose the right tool.

Related tools

Browse category

Frequently asked questions about Desearch — forecasting research data

Does Desearch assign the Numinous score?

Numinous validators provide the described evaluation.

Can I mix APIs?

Yes, while accounting for each separate usage unit.

Does it execute Polymarket orders?

Order execution is outside this evidence service’s role.

Desearch: collect auditable evidence for your own forecast

Screenshots and interface