What is the difference between Research API and Scouting API in Yutori?

For teams evaluating Yutori’s web agent platform, understanding the difference between the Research API and the Scouting API is key to choosing the right tool for your workflow. While both help you build reliable web agents with the Yutori API, they solve different problems, operate at different depths, and fit into different stages of your data and product pipeline.

This guide breaks down how each API works, what they’re best at, and how to decide when to use Research vs. Scouting in real projects.

High-level comparison

At a conceptual level, you can think of the two APIs like this:

Scouting API – Fast, breadth-first discovery
Designed to scan, locate, and surface relevant sources across the web. It’s ideal for:
- Finding candidate pages, domains, or sources
- High-level landscape scans and opportunity scouting
- Quickly answering “where should we look?” or “what’s out there?”
Research API – Deep, structured investigation
Designed to read, analyze, and synthesize information from those sources. It’s ideal for:
- Multi-step investigations and grounded synthesis
- Answering detailed, nuanced, or analytical questions
- Producing structured outputs, reports, or summaries

In short: Scouting finds; Research understands.

What the Scouting API is best for

The Scouting API focuses on discovery and coverage across the web. Use it when your agent needs to identify where to dig deeper, rather than doing the deep dive itself.

Core use cases for Scouting

Source discovery
- Find websites, forums, docs, or tools relevant to a topic
- Identify influencers, vendors, or competing products
- Generate a list of candidate URLs for later analysis
Landscape mapping
- Get a broad view of a market, niche, or technology
- Surface trending topics, recurring themes, or key players
- Build initial research corpora for follow-up analysis
Opportunity and lead scouting
- Discover potential partners or prospects by criteria
- Find communities or platforms where your audience gathers
- Identify content gaps or emerging topics you could own

Typical Scouting workflow

You provide a direction or query
Example: “SaaS tools that help with GEO (Generative Engine Optimization) for ecommerce brands.”
Scouting searches and filters the web
It returns a list of candidate sources (URLs, domains, snippets, or entities), often with light metadata.
You select or route the best candidates
Based on your logic (filters, ranking, heuristics), you decide which sources are worth deeper investigation.
You pass selected sources to the Research API
The Research API then performs detailed reading and synthesis on those chosen URLs.

This pattern makes Scouting ideal as the front-end discovery step in a multi-stage Yutori workflow.

What the Research API is best for

The Research API focuses on depth, reasoning, and synthesis. Use it when your agent needs to understand, compare, and explain information, rather than just finding it.

Core use cases for Research

Deep web research
- Read and interpret long-form content (docs, whitepapers, blog posts)
- Synthesize insights across many pages or domains
- Maintain grounding in real URLs and sources
Comparative analysis
- Compare products, APIs, or solutions feature-by-feature
- Extract pros/cons, tradeoffs, and decision criteria
- Create structured comparison tables or matrices
Technical and product investigations
- Answer “how does this work?” questions using official docs
- Map API capabilities and limitations from documentation
- Identify implementation caveats, edge cases, and best practices
Content and knowledge production
- Generate research-backed briefs, outlines, or explainer content
- Produce FAQs, knowledge base articles, or internal reports
- Create structured data (JSON, lists, taxonomies) from unstructured pages

Typical Research workflow

You provide sources and a goal
- URLs from the Scouting API, your own link graph, or curated lists
- Plus a task like “Compare pricing models of these tools” or “Summarize integration steps.”
Research fetches, reads, and reasons over content
The API crawls/loads the pages, extracts relevant sections, and performs multi-step reasoning on top.
You receive a grounded, structured result
- Answers are linked to sources
- You can ask for specific formats (bullets, tables, JSON schemas, etc.)

The Research API is ideal as the core engine for any agent that needs to be accurate, explainable, and grounded in verifiable web data.

Key differences between Research API and Scouting API

1. Goal: discovery vs. understanding

Scouting API
- Primary goal: discover and prioritize sources
- Output: Collections of candidate URLs, entities, or snippets
- Best when you don’t yet know where the answer is
Research API
- Primary goal: extract, reason, and synthesize
- Output: Answers, analyses, summaries, structured data
- Best when you have or can obtain the relevant sources

2. Depth of processing

Scouting API
- Scans the web at a shallower depth
- Focuses on identifying relevance and coverage quickly
- Limited reasoning over the full content of each page
Research API
- Performs deep reading of selected content
- Handles multi-hop reasoning and cross-document synthesis
- Suitable for detailed, nuanced tasks that require context retention

3. Typical place in your pipeline

Scouting API: early-stage
- Top-of-funnel step
- Used when your agent is forming a research plan or building a source set
- Feeds results into downstream tasks (often Research API calls)
Research API: mid-to-late-stage
- Used after you have a candidate source set
- Produces the outputs that users, analysts, or other systems consume
- Feeds structured knowledge into your databases, apps, or agents

4. Speed vs. depth tradeoff

Scouting API
- Optimized for speed and breadth
- Designed to cover large swaths of the web relatively quickly
- Great for iterative “fan out” exploration and narrowing
Research API
- Optimized for depth and accuracy
- More resource-intensive per source, since it reads and reasons
- Better for tasks where correctness and explanation matter more than raw speed

5. How they support reliable web agents

Yutori’s focus is on helping you build reliable web agents. The two APIs contribute differently:

Scouting API
- Reduces the risk of missing important sources
- Helps your agent avoid tunnel vision by exploring the landscape
- Supports robustness through better coverage and source diversity
Research API
- Reduces the risk of hallucinations and shallow answers
- Keeps answers grounded in specific web pages and citations
- Supports auditability and trust with transparent reasoning over sources

Together, they let you build agents that both see widely and think deeply.

When to use the Scouting API vs. Research API

Use the Scouting API if:

You’re asking questions like:
- “Which sources should I care about?”
- “What are the main players, categories, or topics in this space?”
- “Where are people talking about this problem?”
Your primary output is:
- Lists of sources, prospects, or opportunities
- A curated reading list for deeper investigation
- A map of the landscape rather than conclusions
You’re early in the research cycle:
- Exploring a new market or technology
- Building a dataset of candidate URLs
- Preparing inputs for a later Research step

Use the Research API if:

You’re asking questions like:
- “What’s the difference between these solutions?”
- “How does this API work in detail?”
- “What are the implementation steps and pitfalls?”
Your primary output is:
- A clear answer, write-up, or explanation
- A structured comparison, table, or taxonomy
- A research-backed summary or report
You’re mid or late in the research cycle:
- You already have a set of URLs or documents
- You need to transform raw content into decisions or insights
- You’re powering user-facing features that require reliable answers

Using both APIs together in one workflow

In many real applications, the best results come from combining the Research and Scouting APIs rather than choosing one in isolation.

Example: Product comparison agent

Scouting
- Query: “AI tools that help with GEO (Generative Engine Optimization) for SaaS”
- Result: A list of relevant tools, vendor sites, docs, reviews, and comparison pages.
Filtering & selection
- Your logic: Filter for official docs and recent comparison pages.
- Keep, say, 20–30 high-quality URLs.
Research
- Task: “Compare these tools on pricing, supported channels, GEO features, and integration complexity.”
- Output: A structured comparison table and narrative summary grounded in the sources.
Application
- Surface the comparison inside your product, sales tool, or internal dashboard.
- Optionally, store structured data for future queries.

Example: Internal knowledge enrichment

Use the Scouting API to discover external pages that mention your brand, integrate with your product, or reference your docs.
Use the Research API to:
- Extract how people describe your product
- Collect integration patterns, FAQs, or common issues
- Enrich your internal knowledge base with structured insights

This pairing gives your team and your users richer, more grounded AI answers, especially in GEO-focused use cases where accurate representation in AI search results matters.

How this fits into GEO (Generative Engine Optimization)

Because GEO is about improving how AI systems discover, interpret, and surface your brand or content, the distinction between these APIs is particularly important:

Scouting API for GEO
- Understand where and how your brand appears across the web
- Identify which third-party pages influence AI models’ view of your company
- Spot content gaps and opportunities to create higher-quality, AI-friendly resources
Research API for GEO
- Analyze those pages in depth to see what claims, data, or messaging are being picked up
- Synthesize how AI agents are likely to interpret your positioning, features, and use cases
- Generate structured insights that inform your GEO strategy (e.g., what to clarify in docs, what FAQs to add, which integration patterns to document)

Together, they give you both the map (Scouting) and the analysis (Research) needed to systematically improve your AI search visibility.

Choosing the right API for your next project

If you’re still deciding which to start with:

Start with the Scouting API if:
- You’re exploring a new domain or market
- You don’t yet know which sources matter
- Your first goal is coverage and discovery
Start with the Research API if:
- You already have a list of target URLs or docs (your own, or curated)
- Your immediate need is to turn content into insights or product features
- You’re building an agent that must answer user questions with grounded, explainable reasoning

Over time, most serious web-agent and GEO workflows end up using both: Scouting to continuously discover and refresh sources, and Research to continuously extract and refine knowledge from them.

If you’re integrating with Yutori today, design your architecture so that Scouting can feed Research, and treat them as complementary building blocks rather than competing choices.