VantegrateNative API integrationIntegrate Looker with AI agentsusing the metrics in your LookML model
Metrix, Vantegrate's data agent, connects to your Looker instance through the official Looker API: your team asks on WhatsApp, Slack or Microsoft Teams, the agent builds the query against your Explores, and Looker generates the SQL from your model's definitions. The number matches your dashboards, and the data stays in your database.
How the data flows
What does it mean to integrate Looker with Vantegrate's AI agents?
Integrating Looker with Vantegrate's AI agents means that Metrix, the data agent, queries your Explores through the official Looker API: it turns a plain-English question into the fields, filters and measures of your LookML model, and Looker generates and runs the SQL against your database. The answer comes back on WhatsApp, Slack or Microsoft Teams, with the same numbers your dashboards show.
Looker isn't the same as Data Studio, Google's no-cost reporting tool that used to be called Looker Studio. Google positions Looker as its enterprise BI platform, built on a central semantic model, and Data Studio as the place for ad hoc reports. If your dashboards live in Data Studio, Metrix connects to the sources that feed them, such as BigQuery or Google Sheets.
What do you want to connect to your Looker?
Pick what you need and we'll write the message for you.
1 · Pick what to solve
2 · What our team receives
Vantegrate
online
What data Metrix queries in your Looker instance
The agent works on what your data team has already modeled in Looker, in read-only mode: it doesn't copy tables or recalculate metrics on its own.
Explores, dimensions and measures
The agent reads the structure of each enabled Explore (its dimensions, measures and descriptions) and builds the query from those fields, without hand-writing SQL.
Saved Looks
When a Look already answers the question, the agent can run that saved query and return the current result, with the filters that were set in Looker.
The sources behind your Data Studio reports
For what lives in Data Studio, the agent goes to the report's source (BigQuery, Google Sheets or Google Analytics 4) with its own read-only access.
How the integration works, step by step
Vantegrate handles the setup and rollout. Your Looker admin authorizes it, your data team validates it, and everyone else starts asking.
A least-privilege API user
Your admin creates the API credentials (client ID and client secret) on a dedicated user, with a role that only sees the models and Explores in scope, as Google recommends.
Enabled Explores and synonyms
We choose which Explores the agent can query and add the words your team actually uses to Metrix's semantic layer: region, channel, product family.
From question to Looker query
Metrix turns each question into a query on the Explore (fields, filters and sorting) and runs it through the API; Looker generates the SQL from your LookML and returns the result.
Validation against your dashboards
We compare the answers with the dashboards your team already uses, fix whatever doesn't match and roll out one department at a time.
Looker vs. Data Studio (formerly Looker Studio): what's the difference
They are two Google Cloud products that shared a name. Data Studio is the no-cost tool for building reports and dashboards with drag and drop, with a paid edition, Data Studio Pro. For a few years it was called Looker Studio, until Google brought back its original name (Google Cloud). Existing reports keep working: lookerstudio.google.com redirects to the new domain.
Looker is the enterprise BI platform. Your data team defines dimensions, measures and table joins once, in LookML, and Looker uses that model to construct the SQL queries against your database (Google Cloud). Google reserves it for use cases that need data governed by a central semantic model, and points to Data Studio for personal exploration and ad hoc reports.
| Aspect | Looker | Data Studio (formerly Looker Studio) |
|---|---|---|
| What Google positions it for | Enterprise BI with a central semantic model | Ad hoc reports and personal exploration |
| Cost | Platform plus per-user licenses, on an annual subscription | No cost; Data Studio Pro is the paid edition |
| Metric definitions | LookML: dimensions, measures and joins defined once | The fields each report data source configures |
| API | REST API 4.0: queries, content, users and scheduled deliveries | Data Studio API: search and manage reports and data sources |
| How Metrix uses it | Queries your Explores through the API, with your model's metrics | Connects to the report's sources, not to the report |
Based on Google Cloud documentation and blog posts. The Data Studio API is available only to organizations with Google Workspace or Cloud Identity.
If you use both: Data Studio can read Looker Explores through its Looker connector, which a Looker admin enables (Google Cloud). If your reports work that way, Metrix queries the same Explores through the API and answers with the same metrics.
Why the agent asks Looker instead of querying your database directly
An agent that writes SQL against the database has to guess which column holds net sales and how the tables join. Looker has already solved that: the LookML model stores every definition, and Looker's SQL generator applies it to every query. Metrix reads the structure of your Explores and asks Looker for fields, filters and sorting, not SQL; Looker builds the query and runs it (Google Cloud). It's the same semantic layer approach Metrix takes with any source.
The second reason is permissions. In Looker, API credentials are always bound to a user, and every call returns only what that user is allowed to see. Google advises against using admin accounts in production and recommends minimal-privilege users created for API work (Google Cloud). On top of that, Metrix applies the scope of each person who asks: a regional manager sees their region and not the others.
Two facts from Google to keep in mind before you start. In Looker (Google Cloud core), API access comes with the Developer User license, not with the Standard or Viewer licenses, and each platform includes two of them (Google Cloud). On top of that, calls that run queries count against a monthly limit per edition: up to 1,000 on Standard, 100,000 on Enterprise and 500,000 on Embed (Google Cloud). That's why we size the expected question volume with your team before opening the agent to the whole company.
What gets logged: your admin can pull a list of every API call in a given period from Looker's System Activity (Google Cloud), and Metrix also records who asked what and which query ran.
Looker's MCP server, Gemini and where Metrix fits
Google offers a Looker-managed MCP server, built into Looker and in preview, to connect assistants such as Gemini CLI, Claude or Cursor to a Looker-hosted instance through the Model Context Protocol. It uses OAuth 2.1, the assistant inherits the roles of whoever authorizes it, every tool stays off until an admin turns it on, and each action is logged in System Activity (Google Cloud). For customer-hosted instances there's MCP Toolbox for Databases, which is open source and not a supported Google Cloud product.
Looker also ships its own conversational BI powered by Gemini: you can ask an Explore questions in plain English, or ask a data agent that queries up to five Explores and can be shared or published to Gemini Enterprise (Google Cloud). If your team works inside Looker, that may be all you need.
Metrix fits somewhere else: it takes the question to the channel where people already are, whether that's WhatsApp, Slack or Microsoft Teams, and combines Looker with sources that aren't modeled there, such as the ERP, the CRM or a spreadsheet, with one definition per metric. It doesn't replace the MCP server or Gemini, which an analyst can keep using from their own assistant, or Looker's scheduled deliveries, which send a Look or a dashboard by email, webhook, Amazon S3 or SFTP, or to services such as Slack, Microsoft Teams and Twilio through the Action Hub, where WhatsApp isn't listed (Google Cloud).
To be clear: Vantegrate is not a Google partner. Metrix connects through the official Looker REST API, with the user and permissions your admin defines.
You've seen how it connects. Want to walk through your case?
Tell us how your Looker is set up and we'll tell you what data we need and where the agent connects.
An agent connected to your Looker vs. a standalone tool
| Standalone AI tool | Agent connected to your Looker | |
|---|---|---|
| Data source | A copy that drifts | Your Looker Explores, live |
| Metric definitions | Recalculated separately, may not match | Your LookML, the same as your dashboards |
| Where your data lives | In a third-party system | In your database: Looker runs every query |
| Permissions and audit | Broad and hard to audit | Least-privilege API user and a full log |
| Reaching production | Often stalls as a pilot | Validated, phased rollout |
What to know before connecting an agent to Looker
Facts Google publishes about its own products, each with its source. None of them is a Vantegrate result.
1,000
Query-related API calls per month included in the Looker (Google Cloud core) Standard edition; Enterprise goes up to 100,000
1,400+
Data sources Data Studio connects to directly, including Google Sheets, BigQuery and Google Analytics
0
Tools enabled by default on the Looker-managed MCP server: the admin chooses which ones to turn on
5
Explores, at most, per data agent in Looker's built-in conversational analytics
These come from Google Cloud documentation and can change with each release. The API limit is the one for Looker (Google Cloud core) and doesn't count calls from the Data Studio connector; the managed MCP server is in preview, and its calls do count.
Which agent runs on your Looker instance
Looker is a conversational BI use case: the data agent queries your Explores live and answers with the metrics your team has already defined.
The integration doesn't take your data anywhere
The principle is simple: AI comes to your data, not your data to the AI. Metrix queries your Looker instance through the official API, with a least-privilege user, and Looker runs every query against your database.
- Connection through the Looker REST API 4.0 over HTTPS, with the credentials (client ID and client secret) of a dedicated user your admin creates.
- The agent only reads: it queries Explores and Looks, and doesn't edit your LookML, your dashboards or your data.
- Every call is logged in Looker's System Activity and in Metrix's own log, with who asked and which query ran.
- The agents run on Salesforce or Oracle Cloud Infrastructure, whose SOC 2 and ISO 27001 certifications belong to those platforms, not to Vantegrate or Looker.
Frequently asked questions about the Looker integration
What data and IT teams usually ask before connecting an AI agent to Looker or to their Data Studio reports.
Are Looker and Looker Studio the same thing?
Are Looker and Looker Studio the same thing?
No. Looker Studio went back to its original name, Data Studio: it's Google's no-cost tool for ad hoc reports and dashboards, with a paid edition, Data Studio Pro. Looker is the enterprise BI platform, with a semantic model written in LookML. Metrix queries Looker through its API and, for Data Studio, connects to the sources behind your reports, such as BigQuery.
How does Metrix connect to Looker?
How does Metrix connect to Looker?
Through the Looker REST API 4.0. Your admin creates API credentials (client ID and client secret) on a least-privilege user; with them, the agent gets an access token and queries the enabled Explores. Nothing gets installed in Looker and the agent doesn't need direct access to your database: Looker generates the SQL from your model and runs it. In Looker (Google Cloud core), each query counts against your edition's monthly API limit. Learn more about the agent on the Metrix page.
Does Looker have an official MCP server?
Does Looker have an official MCP server?
Yes, in preview. The Looker-managed MCP server works on Looker-hosted instances, connects over OAuth 2.1 and inherits the permissions of whoever authorizes it. Its tools stay off until an admin enables them, it comes at no additional cost, and its calls use the instance's API quota (Google Cloud). It's built for each analyst's own assistant.
If Looker already has conversational analytics with Gemini, why add Metrix?
If Looker already has conversational analytics with Gemini, why add Metrix?
If your team works inside Looker, conversational analytics with Gemini may be enough. Metrix adds two things: it answers on WhatsApp, Slack or Microsoft Teams, where the person asking already is, and it combines Looker with sources that aren't modeled there, such as the ERP or the CRM, with one definition per metric. It works alongside Looker: it doesn't replace your dashboards or your data team's work.
Metrix, the data agent
How your team asks about its KPIs in plain English and gets the answer where it already works.
Learn moreSnowflake integration
How Metrix queries a cloud data warehouse in plain English, with a read-only role.
Learn moreBI Bottleneck Calculator
What every "I'll send you the report tomorrow" costs your data team and your managers.
Learn moreLet's connect AI agents to your Looker instance
Tell us which Explores and dashboards your team uses (and whether you also have reports in Data Studio), and we'll show you what the agent queries, which user and permissions we request and what it looks like to ask about your KPIs on WhatsApp. A 30-minute conversation, no commitment.
Francisco Morales, co-founder, takes your call. We reply on WhatsApp within 4 business hours, no strings attached.
