ZoomInfo MCP and the ZoomInfo API are now available in Snowflake, the AI Data Cloud. Revenue, data, and analytics teams can bring ZoomInfo's verified company, contact, intent, and signal data into Snowflake CoWork, Cortex Agents, and their own warehouse pipelines. Both reach Snowflake through GTM.AI, ZoomInfo's headless GTM context layer. Business users ask questions in natural language through the ZoomInfo MCP connector in Snowflake. Data teams enrich tables in real time or in bulk through the API.
What is the ZoomInfo Snowflake integration?
The ZoomInfo Snowflake integration gives customers two new ways to reach ZoomInfo's B2B data from inside Snowflake. The first is ZoomInfo MCP, a Model Context Protocol server that Snowflake CoWork and Cortex Agents call as a tool. The second is the ZoomInfo API, which data engineers call from Snowflake Tasks, Streams, Snowpark pipelines, or external functions.
Both sit on top of an integration Snowflake customers already use. ZoomInfo's Data Cubes on Snowflake Marketplace, part of ZoomInfo Data as a Service, deliver pre-built company, contact, intent, hierarchy, and IP datasets into a customer's warehouse on a refresh schedule. The Data Cubes give a team a queryable baseline. The ZoomInfo API adds on-demand and transactional access. ZoomInfo MCP adds a conversational layer that business users reach without writing SQL. For a deeper comparison of the two access patterns, see MCP vs. API for data enrichment.
How does ZoomInfo MCP work in Snowflake CoWork?
Snowflake CoWork, formerly Snowflake Intelligence, is Snowflake's AI work agent. It runs on Cortex Agents, the framework that combines Cortex Analyst (text-to-SQL), Cortex Search, custom tools, and MCP connectors. ZoomInfo's MCP server plugs into that framework as an external MCP server.
Setup follows Snowflake's standard MCP connector flow. A Snowflake admin creates an API integration object holding the ZoomInfo server URL and OAuth details. The admin then creates an External MCP Server object and adds it to a Cortex Agent. End users open CoWork, go to Sources, then Connectors, and authenticate to ZoomInfo with their own ZoomInfo login.
Once the connector is live, a CoWork agent can call 14 ZoomInfo MCP tools across three groups:
Finding: Search Companies, Search Contacts, Find Similar Companies, Find Similar Contacts, Find Recommended Contacts, Search Intent, Search Scoops
Enriching: Enrich Companies, Enrich Contacts, Enrich Intent, Enrich Scoops, Enrich News
Researching: Account Research, Contact Research
Each answer combines the customer's governed data in Snowflake with live ZoomInfo context, so a question about pipeline, territory, or account health returns internal and external facts together.
How does the ZoomInfo API work with Snowflake?
The ZoomInfo API is for teams that need programmatic, high-volume, or event-driven access instead of a conversation. It returns JSON and is organized into three tiers:
Tier | APIs | Best for |
|---|---|---|
Standard | Search, Enrich, Lookup, Usage | Real-time, record-by-record lookups and enrichment |
Scaling | Bulk, Monitoring (Webhooks) | Warehouse-scale backfills and change tracking |
Specialized | Contact, Company, Company Master, Org Chart Enrich, Chorus | Org structure, location-level company data, conversation intelligence |
The Standard API supports 25 requests per second, and the Scaling API supports 33 requests per second.
What can teams do with ZoomInfo in Snowflake?
The use cases below show how sales, RevOps, data engineering, data science, and marketing teams split the work between ZoomInfo MCP and the ZoomInfo API.
1. Account research inside Snowflake CoWork (MCP)
An account executive preparing for a renewal asks CoWork: "Summarize the health of our top 20 accounts in the Northeast and flag any with leadership changes in the last 90 days."
Cortex Analyst pulls usage, ARR, and support tickets from the customer's Snowflake tables. The agent then calls ZoomInfo and Account Research tools to add executive moves, funding events, and headcount trends. The rep gets one cited answer that combines internal account history with what is changing at each company, without exporting a list or opening another tool.
2. Buying-committee mapping for open opportunities (MCP)
A sales manager wants to map the buying committee on every large deal, and asks: "For every open opportunity over $100K, who else should we be talking to?"
The agent reads open opportunities from the warehouse, then calls Find Recommended Contacts and Enrich Contacts to return verified titles, seniority, and contact details for missing stakeholders. Because ZoomInfo MCP respects the user's existing ZoomInfo permissions, reps only see contact data their ZoomInfo seat already entitles them to.
3. Real-time lead enrichment as records land (API)
A RevOps team runs lead enrichment on inbound form fills as they land in a Snowflake staging table. A Snowflake Stream detects each new row, and a Task calls the ZoomInfo enrich API through an external function. Firmographic and technographic and contact fields come back within the same call. Lead scoring and routing models that run downstream in Snowflake always use complete, current records instead of whatever the prospect typed into the form.
4. Nightly bulk enrichment and CRM backfill (API)
A data engineering team schedules a nightly Bulk API job against net-new accounts loaded from Salesforce or HubSpot into Snowflake. The Bulk API enriches up to 10,000 records per request and retrieves up to 200,000 records per query, so a full backfill of a mid-market CRM fits in one scheduled run. Revenue, employee count, industry, and tech stack fields stay current in the warehouse without manual exports, which is the same CRM data enrichment job most teams run by hand today.
5. Intent and Scoops signals in propensity models (API)
A data science team pulls ZoomInfo intent data (buying signals across thousands of topics) and Scoops (leadership changes and strategic moves) into feature tables in Snowflake. Those features feed propensity-to-buy and account-scoring models built in Snowpark, an approach covered in more depth in advanced data scoring in Snowflake. Scores update when signals change, and the same scored accounts are available to CoWork agents for follow-up questions.
6. Change detection with the Monitoring API (API)
A marketing operations team subscribes to the Monitoring API (Webhooks) for previously purchased company and contact records. When ZoomInfo updates a record, such as a contact changing jobs or a company relocating, a webhook fires and a Snowflake Task refreshes only the affected rows. Segments and suppression lists stay accurate without re-pulling the full dataset.
7. Territory and hierarchy planning (Data Cubes + MCP)
A sales operations team uses the ZoomInfo Hierarchy and Company Data Cubes already in Snowflake to build parent-child account trees for an annual sales territory plan.
During planning, leaders ask CoWork follow-up questions such as "Which subsidiaries of our top 50 parents have grown headcount more than 20% this year?" The agent answers from the Data Cubes and fills gaps through ZoomInfo MCP. The Data Cubes give the stable baseline, and MCP adds live context on demand.
Why does GTM.AI matter for Snowflake customers?
GTM.AI is a headless GTM context layer. It exposes ZoomInfo's verified data graph and agentic orchestration through API and MCP, so any agent, platform, or workflow can plug in. The Snowflake integration is one of dozens built on it.
The data backbone is the GTM Context Graph: an identity-resolved graph of 100M companies, 500M contacts, and billions of signals that is continuously updated and continuously queryable. Above the graph, agentic orchestration runs agent teams that handle enrichment, research, outreach, and execution, using proprietary GTM models, tools, and skills. The Account Research and Contact Research tools available in Snowflake CoWork are direct outputs of that orchestration layer.
For Snowflake customers, the result is bidirectional. The customer's first-party data and business context live in Snowflake. GTM.AI supplies the external market view that first-party data cannot provide on its own. Agents and pipelines in Snowflake reason over both.
How is ZoomInfo data governed inside Snowflake?
GTM.AI applies the same governance, access control, permissioning, data lineage, and audit logging on every surface that consumes it. Inside Snowflake, that means:
Authentication: OAuth 2.0 for both ZoomInfo MCP and the ZoomInfo API, with PKCE on the API.
Permissions: Every call inherits the user's existing ZoomInfo role-based permissions.
Admin control: Only a Snowflake admin (or an explicitly granted role) can create the API integration and External MCP Server objects that expose ZoomInfo to Cortex Agents.
Auditability: All ZoomInfo usage is logged in ZoomInfo's auditing framework and visible through the Usage API.
No permanent storage by the agent: AI assistants do not permanently store ZoomInfo data returned through MCP.
Snowflake's external MCP connectors currently support tool calls only. Resources, prompts, and sampling are not supported, so ZoomInfo MCP in Snowflake is scoped to answering questions and running tools on demand. Bulk delivery stays with the API and Data Cubes.
How does Snowflake fit into the GTM.AI ecosystem?
The Snowflake integration is one of many ways customers reach GTM.AI. The same headless context layer powers pre-configured agents in Salesforce, HubSpot, Microsoft, Gong's profile iFrames, and customer-built agents on Claude, ChatGPT, and Glean. Developer and automation surfaces including Replit, Cursor, and n8n reach it as well. Every surface uses the same data graph and the same governance.
Snowflake customers can review setup options and implementation guides on the ZoomInfo Snowflake integration page.
FAQ: ZoomInfo on Snowflake
What is the ZoomInfo MCP connector for Snowflake?
The ZoomInfo MCP connector for Snowflake links Snowflake CoWork and Cortex Agents to ZoomInfo MCP, the Model Context Protocol server for GTM.AI. Snowflake users can search, enrich, and research companies and contacts in natural language, alongside their own data in Snowflake.
Is ZoomInfo MCP the same as the ZoomInfo Data Cubes on Snowflake Marketplace?
No. Data Cubes deliver pre-built company, contact, intent, hierarchy, and IP datasets into a customer's Snowflake account on a refresh schedule. ZoomInfo MCP answers questions on demand inside CoWork and Cortex Agents. The two are complementary.
When should a team use the ZoomInfo API instead of MCP in Snowflake?
Use the ZoomInfo API for programmatic, event-driven, or high-volume work such as real-time enrichment from Snowflake Tasks, nightly bulk backfills, and webhook-driven refreshes. Use ZoomInfo MCP for interactive questions from business users in Snowflake CoWork.
What does a Snowflake admin need to set up ZoomInfo MCP?
A Snowflake admin needs privileges to create an API integration and an External MCP Server object, plus Cortex Agents or Snowflake CoWork enabled on the account. Each end user then authenticates to ZoomInfo through OAuth from CoWork's Connectors menu.
Does ZoomInfo MCP in Snowflake respect existing ZoomInfo permissions?
Yes. Every ZoomInfo MCP and ZoomInfo API call inherits the user's existing ZoomInfo role-based permissions. All usage is logged in ZoomInfo's audit framework.
How does the Snowflake integration relate to GTM.AI?
GTM.AI is ZoomInfo's headless GTM context layer. ZoomInfo MCP and the ZoomInfo API are the two interfaces through which GTM.AI reaches Snowflake, using the same GTM Context Graph and governance that power agents in Salesforce, HubSpot, Microsoft, Claude, and ChatGPT.
Start building with GTM.AI
GTM.AI gives your agents and pipelines direct access to ZoomInfo's GTM Context Graph through API and MCP. Start building with the same data and governance that power ZoomInfo's Snowflake integration.
