Search and web retrieval
How to use
Tavily.
Giving AI applications current web search, clean page extraction, selective site crawling, URL mapping, and asynchronous research capabilities.
Tavily is a web search, extraction, crawling, mapping, and research API for AI applications. Giving AI applications current web search, clean page extraction, selective site crawling, URL mapping, and asynchronous research capabilities. This guide covers the whole path in one place: official access, a first session that produces something reviewable, the checks that make output trustworthy, and the permissions worth limiting before you connect real work.
AI should make the work easier to inspect. If the workflow removes the source, the owner, or the review step, redesign the workflow.
Access & setup
Find, install, and sign in to Tavily
Get into the official Tavily experience with the right account and a setup you understand.
- 01
Start at https://www.tavily.com/ and confirm the domain before entering account or payment information.
- 02
Availability: Tavily is accessed through its web platform, playground, HTTP APIs, and Python or JavaScript SDKs.
- 03
A Tavily account and API key, network access, and a secure application or supported development environment.
- 04
Sign in with the account you intend to keep using, then review plan, data, notification, and permission settings.
- 05
Run one low-risk test task before connecting sensitive files, repositories, or workspace data.
- Use official download pages.
- Review permissions during setup.
- Keep installers and applications updated.
First session
Your first useful Tavily session
Learn the interaction loop using a small task with a clear outcome.
- 01
Create an API key and keep it outside source code and browser-delivered applications.
- 02
Define a representative search need with topic, freshness, source, depth, and result-count requirements.
- 03
Run a basic filtered search and inspect returned URLs and content before enabling generated answers or deeper retrieval.
- 04
Track usage by project and test relevance, citations, rate limits, failures, and credit cost before integration.
- State the outcome before the background.
- Provide the real source material.
- Review the result before expanding the task.
Quality control
Check the quality of Tavily output
Establish that a retrieved answer is actually supported by the sources it cites.
- 01
Restate the question precisely; retrieval quality degrades fastest on vague or compound queries.
- 02
Open the cited sources and confirm each one genuinely contains the claim attributed to it.
- 03
Check the publication and capture dates — stale sources are the most common cause of confidently wrong answers.
- 04
Look for what is missing: retrieval returns what matches the query, not what would contradict it.
- 05
Confirm quotations are verbatim and that summaries have not reversed a qualification or a negation.
- Open the citation; do not trust that a link supports the claim beside it.
- Check dates on every retrieved source.
- Treat an absent contradiction as unverified, not as agreement.
Privacy & permissions
Use Tavily safely
Control what you index and what leaves your systems when you query it.
- 01
Classify the corpus before indexing: confidential, personal, and regulated material needs an explicit decision.
- 02
Confirm where the index is stored, who can query it, and how long content is retained.
- 03
Scope crawling and scraping to sources you are permitted to collect, and respect robots and terms of use.
- 04
Strip secrets and personal data before indexing rather than filtering them at query time.
- 05
Review query logs — the questions asked can be as sensitive as the documents indexed.
- Retrieved and generated content may be stale, irrelevant, unsafe, or unsupported. Protect keys, constrain domains and crawls, preserve sources, treat web text as untrusted, verify consequential claims, respect site and data rights, and monitor credits and limits.
- Follow your organisation's approved-use policy.
- Never treat fluent output as authorization to act.
Core workflows
Step-by-step ways to use Tavily for the work it does best.
Each workflow below has its own outcome, steps and review checkpoints.
Tune search depth, topic, date, domains, and response size for evidence a model can inspect.
↓ 02 WorkflowCreate a bounded website crawl with TavilyDiscover and extract only the relevant parts of an authorized site for documentation or retrieval ingestion.
↓Workflow 01
Build a citation-ready search workflow with Tavily
Tune search depth, topic, date, domains, and response size for evidence a model can inspect.
Outcome: Retrieve timely, relevant sources without allowing a generated search answer to replace source verification.
- 01
Define the query intent, topic, freshness, geography, acceptable domains, maximum sources, and evidence required for an answer.
- 02
Choose basic, fast, ultra-fast, or advanced depth deliberately and set date and domain filters instead of relying on defaults.
- 03
Request raw content or generated answers only when justified by the context and credit budget, preserving every source URL.
- 04
Evaluate representative and adversarial queries for ranking, freshness, duplicate sources, unsupported answers, and missing coverage.
- 05
Pass selected excerpts with citations to the model and log project, query settings, request outcome, credits, and source count safely.
What good use looks like
- Choose depth by measured need.
- Verify generated answers against results.
- Track usage by project.
Workflow 02
Create a bounded website crawl with Tavily
Discover and extract only the relevant parts of an authorized site for documentation or retrieval ingestion.
Outcome: Collect a controlled, auditable corpus without crawling unnecessary paths or overflowing cost and context limits.
- 01
Confirm authorization and define the root domain, allowed paths, excluded private areas, freshness need, page limit, and update cadence.
- 02
Use Map when only URLs are needed or Crawl for content, starting at depth one with conservative breadth and page limits.
- 03
Restrict domains and paths, disable external traversal unless required, and use instructions or chunks only for a clear semantic target.
- 04
Validate URLs, content, duplicates, redirects, extraction errors, prompt injection, personal data, and page ownership before indexing.
- 05
Store source URL, retrieval time, hash, and access policy with each document, then test incremental refresh and deletion handling.
What good use looks like
- Map before a large crawl.
- Start with depth one.
- Keep access and lineage metadata with every page.
Official references
Check the current product documentation.
Features, plan limits, availability, and data controls change. These official pages are the starting points used for this guide.