What search provider do you use?

We rotate between providers (Brave, SerpAPI, Tavily) for resilience and quality. Results are normalized to a single schema before the model sees them.

Can it read sites behind login?

Not Fetch — that uses an unauthenticated HTTP client. For login-gated sites, use Browser Control, which spins up a real Chromium with your saved session.

How fresh is the data?

Search results are real-time. Fetched pages are loaded live, not cached, so what you see is what's currently on the page.

What about paywalled sites?

Fetch reads what the publisher serves; some paywalled sites only return a teaser. We don't bypass paywalls. For research-grade access, point at open-access mirrors or use Deep Research with academic source pinning.

Tools & Automation

Web Search and Fetch

Search the open web and read any URL, with citations.

Two foundational tools every chat has access to. Web Search runs a real search query and returns ranked results. Fetch loads any URL and reads it cleanly. Combined, they let Zentor answer questions about live information — news, prices, docs, code repos — without leaving the chat.

Get started All integrations

How it works

3 steps to wire up Web Search and Fetch, no engineering required.

1

Ask a question that needs the web

'What's the latest pricing for AWS RDS?', 'Read this docs page and summarize it', 'What did they announce at the Apple event today?'
2

Zentor runs search and fetch as needed

It might search to find candidate URLs, fetch a few of them, then synthesize. Citations are inline so you can verify the source of every claim.
3

For multi-source investigations, use Deep Research

Search and Fetch are good for 1 to 3 sources. For 10+ sources or a structured report with cross-checking, Deep Research is the right tool.

Try saying

Real prompts you can paste into Web Search and Fetch.

› What's the current price of a 2TB SSD on Amazon, and is it cheaper than 6 months ago?
› Read the Anthropic API pricing page and tell me the cached vs uncached input rate for Sonnet 5.
› Find the GitHub repo for langchain's Postgres vector store and read the README to me.

Step by step demo

What actually happens when you send the prompt.

Prompt 01 3 steps

“Read the Anthropic API pricing page and tell me the cached vs uncached input rate for Sonnet 5.”

What Zentor does

1 Searches for 'Anthropic API pricing site:anthropic.com'.
2 Fetches the top result (anthropic.com/pricing) and reads the model table.
3 Extracts the Sonnet 5 row: input $3 per 1M tokens, cached input $0.30 per 1M (90% off).

Result

Replies: 'Sonnet 5 is $3 per million input tokens uncached, $0.30 per million when cached — a 90% discount. Cache writes are billed at $3.75 per million (a 25% premium over uncached). Source: anthropic.com/pricing.'

Built for these workflows

Workflows people automate with Web Search and Fetch on Zentor.

Use case