NEW Live web search, scrape and research in one API.

The web API for AI agents.

One API for live web search, extraction, and research. Clean, consistent structured results every run, so your agent never breaks.

live web, structured JSON
SCOUT — SEARCH
~/scout · api v1 · typed json, every call

// WHY SCOUT

Why Scout?

Scout turns the live web into clean, typed JSON. It is the same shape every run, so your agent does not break when a page changes.

SRC · example.com/vector-db-guide200 OK · 280ms

A Practical Guide to Vector Databases

Vector databases have moved from a research curiosity to core infrastructure for retrieval-augmented applications in only a few short years.

At their core, vector databases index high-dimensional embeddings so nearest-neighbour lookups stay fast as the corpus grows, trading exact matches for speed at scale.

Benchmarks published this year show retrieval latency holding flat well past a billion vectors on commodity hardware, which is what finally made them production-ready.

The trade-offs are real: index build time, memory pressure, and the cost of re-embedding whenever the model changes.

01

Live web access for agents

Give an agent a web tool in one call. It searches the live web and reads any page into clean JSON, so it works from current facts instead of stale training data.

SEE DOCS →
02

Research with citations

Ask a question and get a typed answer back, with a citation behind every field. Scout runs the multi-step research a person would do by hand.

SEE DOCS →
03

Grounded answers, no guessing

A drop-in for the OpenAI chat endpoint where every reply is grounded in live search. Change one base URL and your app stops making things up.

SEE DOCS →
04

Web pages into your schema

Point Scout at any URL with a JSON schema and get back exactly those fields. The HTML is handled for you, and the shape stays the same every run.

SEE DOCS →
05

Datasets from a description

Describe what you want, like "Series B fintechs in the EU", and get every match back as typed rows. Then enrich each row with the fields you need.

SEE DOCS →
06

Watch the web for changes

Track a query on a schedule. Scout posts to your webhook the moment results change, so the live web becomes an event stream your product can react to.

SEE DOCS →
EARTH TIME
00:00:00:00
| | | | |   S C O U T  O P S  C O N S O L E  | | | | |
SESSION
00:00:00:00
— RETRIEVAL —
SEARCH[ ████████████▒▒▒▒ ] 78%
EXTRACT[ ██████████▒▒▒▒▒▒ ] 64%
READ[ ███████████████▒ ] 91%
— WORKER POOL —
NODE A[ ████████████████ ] 100%
NODE B[ ████████████████ ] 100%
CACHE HIT RATE
— REGIONS —
IADSFOFRA
[]
REQ 4821
200  OK
280  MS
A U T O
TRACKING  THE  LIVE  WEB
— RESULTS —
LATENCY MS
500475450425400375350325300275250225200175150125100075050025000500475450425400375350325300275250225200175150125100075050025000
— CAPACITY —
TASK[ ██████▒▒▒▒▒▒▒▒▒▒ ] 40%
FINDALL[ ████▒▒▒▒▒▒▒▒▒▒▒▒ ] 25%
MONITOR[ ██████████▒▒▒▒▒▒ ] 60%
— STATUS —
CACHE312 ENTRIES
UPTIME99.98%
QUEUE0
SCOUTAPI v1.0REGION: IAD · SFO · FRA

Pay by the call.

Retrieval is cheap; reasoning costs more. Each endpoint is metered in credits, so you pay for the work a call actually does. Cached responses replay free.

SEE PLANS & COST ESTIMATOR →
  • /search1 credit
  • /monitors2 credits / run
  • /extract2 credits / url
  • /chat12 credits
  • /task45 credits
  • /findall80 credits

Point your agent
at the whole web.

Your agent gets web context it can act on. Scout searches and reads the pages, then returns them as typed JSON. Pay only for the calls you make.

BASE URL · https://core.usescout.sh