Tavily Is Now Available in Arcade.dev: Credible, Cited Research for AI Agents to Act On

The gap between AI reasoning and real-world action is where most agent projects stall. The model can analyze. But can it search the web for breaking news, write that analysis into a Google Doc, and post a summary to your team's Slack channel securely, on behalf of the right user, with a full audit trail?

Today, we're announcing that Tavily's real-time search and research tools are available in Arcade.dev, the MCP runtime for AI agents. This means any team building agents in finance, healthcare, legal, or beyond can now access Tavily's web intelligence capabilities through Arcade's secure, governed infrastructure. No ad-hoc security plumbing. No fragile, one-off integrations. No compliance gaps.

Tavily joins Arcade's catalog of 8,000+ agent-optimized tools across 80+ MCP servers. Developers can compose Tavily's research capabilities alongside Google Docs, Slack, Salesforce, GitHub, and dozens of other services each federated through Arcade's MCP Gateway with authorization, execution reliability, and governance built in.

Why This Matters

Tavily is trusted by over 1.5 million developers and processes more than 150 million API queries per month. Their search, research, and extraction endpoints consistently rank at the top of benchmarks, and are used by companies like Jetbrains, Cohere and BCG for agent-powered workflows. But in enterprise environments, powerful tools aren't enough. Security teams ask:

• Who authorized this agent to search the web?
• Are credentials exposed to the LLM?
• Is there an audit trail for every tool call?

These are the exact questions that Arcade was built to answer. As the MCP runtime, Arcade delivers three things simultaneously: agent authorization, reliable tool execution, and governance so agents can take action across any system with granular permissions and complete visibility.

By bringing Tavily into the Arcade ecosystem, we've collapsed the integration complexity. Developers get Tavily's world-class research capabilities with Arcade's enterprise-grade runtime and they can compose Tavily alongside any combination of Arcade's other tools through purpose-built MCP Gateways.

What You Can Build

The Tavily + Arcade integration opens up use cases that were previously too complex to ship with production-grade security. Here are some of the most common patterns we see across industries:

Financial Research and Pre-Market Briefings

Analysts at banks and investment firms spend hours each morning pulling overnight news, scanning SEC filings, and writing briefings. With Tavily's tavily_research endpoint running through Arcade, an agent can conduct deep multi-source research on portfolio companies, create a formatted briefing in Google Docs, and post a summary to the team's Slack channel all in minutes.

Every step is authorized through Arcade's runtime. The analyst authenticates once. Arcade handles the rest: managing credentials for Google, Slack, and Tavily without ever exposing tokens to the LLM. Every tool call is logged and auditable.

Company Research and Competitive Intelligence

Whether it's preparing for a sales call, evaluating an M&A target, or monitoring a competitor's public announcements, Tavily's search and extract tools can pull real-time data from across the web. Combined with Arcade's document and communication integrations, teams can go from question to published report in a single agent conversation.

CRM Enrichment and Lead Research

Sales teams use Tavily through Arcade to enrich CRM records with real-time company data, recent funding rounds, leadership changes, product launches without manual research. The agent searches, extracts, and updates records through Arcade's Salesforce and HubSpot integrations, with every action governed by the same authorization policies.

Automated Analyst Briefs and Market Monitoring

Research and advisory firms use this combination for automated daily brief generation: structured, summarized research across analyst teams, replacing hours of manual work with governed, auditable agent workflows.

How It Works: Building a Financial Intelligence Agent

To show what this looks like in practice, we built an open-source Financial Intelligence Agent a web-based chat application where an analyst can research companies, create Google Docs briefings, and share insights to Slack, all through one conversation.

The agent uses Tavily for real-time search and deep research, Google Docs for document creation, and Slack for team communication. All three services are federated through a single Arcade MCP Gateway. The entire application is under 200 lines of code because Arcade handles the integration complexity.

Here's how we set it up:

1. Get the Tavily MCP URL. From tavily.com, generate a remote MCP link with your API key.

2. Register Tavily in Arcade. In the Arcade Dashboard, add Tavily as a Remote MCP Server. Arcade discovers all six Tavily tools automatically: tavily_search, tavily_extract, tavily_crawl, tavily_map, tavily_research, and tavily_skill.

4. Connect your agent. Any MCP client Cursor, Claude Desktop, VS Code, or a custom application built with the Vercel AI SDK, LangChain, or OpenAI Agents connects to the gateway URL and immediately has access to every tool in the gateway. Authorization, secrets management, and audit logging are handled by Arcade at runtime.

The result: a production-grade agent with real-time research, document creation, and team communication with zero security integration code in the application itself.

Try it yourself: https://github.com/arcadeai-labs/financial-intelligence

Beyond Finance

The pattern here is general. Tavily provides the web intelligence layer. Arcade provides the runtime authorization, execution, governance. Together, they apply anywhere agents need real-time research combined with secure actions:

• Healthcare: Research drug interactions, extract clinical trial data, generate reports for review boards
• Legal: Monitor regulatory changes, research case law, draft compliance summaries
• Media: Automated news monitoring with real-time search, extraction, and alerts
• Insurance: Market analysis, competitor pricing research, document generation
• Sales and GTM: Prospect research, competitive analysis, CRM enrichment, automated outreach prep

Every one of these use cases requires the same foundation: reliable search, governed tool execution, and auditable agent actions. That's what Tavily and Arcade deliver together.

Get Started

Tavily is available in Arcade today.

1. Sign up for Arcade at arcade.dev
2. Get your Tavily MCP URL at tavily.com
3. Add Tavily to your Arcade project and compose an MCP Gateway with the tools you need
4. Connect your agent and start building

For the full technical walkthrough, see the Financial Intelligence Agent repository.

Tavily is an AI-native search provider focused on agents and RAG use cases, serving 1.5M+ developers with 150M+ monthly API queries. Recently acquired by Nebius, a NASDAQ-listed cloud and AI infrastructure company.

Arcade.dev is the only runtime for MCP, delivering secure agent authorization, reliable tools, and governance for enterprise AI agents at scale. Arcade's catalog includes 8,000+ agent-optimized tools across 80+ MCP servers.