Entity and account enrichment
Research a supplied company or prospect list and return consistent fields with supporting sources.
Independent tool overview
Exa Agent at a glance
Exa Agent is a developer API for asynchronous, multi-step web research that can return cited prose or structured data for list-building, enrichment and open-ended investigations.
Visit the official Exa Agent site ↗
- Product type
- Asynchronous web-research API
- Status
- Beta
- Core outputs
- Cited text or schema-matched JSON
- Fixed run price
- $0.012–$1.00
- Default effort
- Auto
- Default concurrency
- Two active runs on standard pay-as-you-go accounts
Overview
What Exa Agent is
Exa Agent combines language models with Exa's search and page-retrieval infrastructure. A developer submits a natural-language research task, optionally supplies input data and an output schema, then receives a long-running agent job that searches, reads and synthesizes sources.
The product is aimed at applications rather than individual researchers. Its strongest fit is a workflow that needs repeatable, machine-readable research—such as enriching companies, assembling a market map or producing a cited brief—not a one-off browser answer.
Exa documents the Agent API as beta, so request headers, behavior and pricing may change. Production teams should also validate source quality, handle partial or failed runs and avoid treating a citation as proof that the model interpreted a source correctly.
Use cases
Who Exa Agent is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Market and competitive mapping
Build lists of products, companies, people or events across many web sources.
Cited research inside applications
Add long-running web investigation to an agent, dashboard or internal workflow through one API.
Capabilities
Core Exa Agent features
Multi-step research runs
Searches, reads and reasons across multiple sources instead of returning a single search-results page.
Structured outputs
The outputSchema parameter can request JSON that matches fields defined by the application.
Bring-your-own input data
The input.data parameter can attach records that the agent should research or enrich.
Citations
Research results include source grounding so downstream users can inspect the supporting pages.
Effort controls
Minimal, low, medium, high, x-high and auto modes trade cost, latency and completeness.
Streaming and run lifecycle
Supports server-sent events plus endpoints for creating, reading, listing, cancelling and replaying run events.
Process
How the Exa Agent workflow works
Step 1
Define the research contract
Specify the question, inclusion rules, freshness requirements and exact fields the application needs.
Step 2
Create an asynchronous run
Send the prompt, effort preference, optional input data and optional output schema to the Agent API.
Step 3
Track completion
Stream lifecycle events or poll the run, with retry and cancellation handling for slow or failed jobs.
Step 4
Validate before storage
Check required fields, inspect citations, deduplicate entities and route uncertain or high-impact results for review.
Cost
Exa Agent pricing and free plan
Fixed effort modes cost from $0.012 to $1.00 per run. Auto effort instead scales compute and search usage to the task. Contact enrichment and specialized Exa Connect providers are charged separately.
Minimal
$0.012 per run
Lowest-cost mode for lightweight tasks.
- Fixed effort cap
- Best for very small lookups
- Enrichment charges are additional
Low
$0.025 per run
Narrow factual tasks and short answers.
- Fixed effort cap
- Lower completeness than larger modes
- Enrichment charges are additional
Medium
$0.10 per run
Exa's recommended starting point for standard research.
- Fixed effort cap
- Suitable for routine cited research
- Enrichment charges are additional
High
$0.50 per run
Harder tasks requiring more research and citations.
- Fixed effort cap
- Higher latency and completeness
- Enrichment charges are additional
X-high
$1.00 per run
Highest fixed effort for valuable completeness-sensitive work.
- Fixed effort cap
- Highest standard fixed cost
- Enrichment charges are additional
Auto
Usage based
Dynamically scales research effort and cost.
- Agent Compute Units are $0.10 each
- Search tool calls are $0.005 each
- Email enrichment is $0.02 and phone enrichment is $0.07 per result
Pricing checked . Check current pricing at the source ↗
Assessment
Exa Agent strengths and limitations
Where it stands out
- One API covers searching, reading, synthesis and citations
- Schema-matched output reduces brittle parsing for database workflows
- Input data and parallel research suit list enrichment at scale
- Fixed effort modes make per-run research budgets predictable
What to consider
- The Agent API is documented as beta and may change before a stable release
- Asynchronous runs require job-state, retry, timeout and cancellation handling
- Default pay-as-you-go concurrency can constrain batch throughput
- Citations expose sources but do not guarantee complete or correctly interpreted results
- Web coverage, freshness and accessibility vary by source, and private or blocked information remains unavailable
Compare
Exa Agent alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Agents
Parallel Search
Choose Parallel Search for another agent-oriented web search and research API.
Explore Parallel Search →Consumer
Gemini Deep Research Agent
Choose Gemini Deep Research Agent for Google's long-running research API and ecosystem.
Explore Gemini Deep Research Agent →Data Analysis
Perplexity
Choose Perplexity when researchers need an end-user answer engine rather than an embedded API workflow.
Explore Perplexity →Questions
Exa Agent FAQs
What is Exa Agent?
Exa Agent is an API that performs asynchronous, multi-step web research and returns cited prose or structured output for applications.
How much does Exa Agent cost?
Fixed modes range from $0.012 for Minimal to $1.00 for X-high per run. Auto effort is billed from Agent Compute Units and search tool calls, with contact enrichment charged separately.
How is Exa Agent different from the Exa Search API?
Search returns web results and content for an application to process. Agent performs a longer workflow that searches multiple times, reads sources, reasons over them and produces a synthesized result.
Can Exa Agent return JSON?
Yes. Developers can supply an outputSchema so the final result follows defined fields, which is useful for enrichment and database pipelines.
Is Exa Agent ready for production?
It can be integrated today, but Exa's documentation labels the API beta. Production use should isolate the integration, monitor changes and validate results before automated decisions.
Bottom line
Our Exa Agent verdict
Exa Agent is compelling for developers who need cited, structured web research inside a product without building their own search-and-synthesis loop. The biggest tradeoffs are beta stability, asynchronous integration work and the need to validate research before it drives a database or decision.
Visit Exa Agent website ↗Similar Agents Tools
View all Agents AI tools →
Poetic
Poetic - AI that learns your procedures, then self-heals when apps change
Voice Agent Builder
Voice Agent Builder- xAI's no-code platform for voice agents running on Grok Voice
Siri AI
Siri AI - Apple's newly overhauled on-device AI assistant
Gemini Managed Agents
Gemini Managed Agents - Google's sandboxed agent API with background tasks and remote tools
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.