Connect Your Own AI to Your Analytics Data

You already pay for an AI assistant. Maybe it’s Claude Pro, maybe Claude Desktop sits in your dock all day. So why should asking it about your own analytics data require a separate API key and a separate bill?

It doesn’t. AI Data Stream now gives every team member a personal connection URL for MCP - the open standard that lets AI assistants securely use external tools. Paste that one URL into claude.ai or Claude Desktop, and your assistant can query the Google Analytics, Search Console, Google Ads, CRM data, annotations, and knowledge base of every property you have access to, across all your teams. In any chat. Using the subscription you already have.

What you get

Your data, in the AI you already use. Every data tool our built-in chat uses is available through the connector: traffic reports, search rankings, ad performance, pipeline summaries, Core Web Vitals, your annotations and knowledge base. Ask “why did traffic drop last Tuesday?” in claude.ai and it pulls the real numbers.

One URL, all your properties. The connector is personal, not per-property. Your AI can list the properties you can access, sees which data sources each one has connected, and knows your permission level on each - so “compare signups across my two sites” just works. Join a new team and its properties appear; leave, and they’re gone.

Your skills come with it. Skills you create in AI Data Stream - from our catalog or your own - travel with the connector. They appear in your AI client’s prompts menu, and your assistant can also discover and run them itself: ask for “the weekly traffic monitor” and it fetches your skill and follows it exactly as you wrote it.

It can take notes, too. The connector isn’t just read access. Spotted the cause of a traffic dip while chatting in Claude? Your AI can record an annotation on the property - attributed to you, visible to your team in the app, and only where you have write permission. It can also save team memories, so facts you establish in one conversation carry into the next.

No API key, no per-token bills. This is the part we like most. Our BYOK philosophy has always been “we don’t mark up AI” - and the MCP connector takes it one step further. If you use claude.ai or Claude Desktop, usage is covered by your existing Claude subscription. Flat rate. No API key to create, fund, or paste anywhere.

Locked down by default. External AI access is off until a team owner explicitly switches it on, and everything else is read-only: the connector can’t touch your Google accounts or change anything in your connected platforms. Your URL is yours alone - regenerate it at any moment and every copy stops working, without affecting your teammates. Owners keep a kill switch: turning the team setting off instantly removes that team’s properties from every member’s connector.

How to connect

1. A team owner enables External AI Access in Team Settings (Pro and Business plans).
2. Every team member can then get their personal URL from Settings → Connected AI.
3. In claude.ai or Claude Desktop: Settings → Connectors → Add custom connector, paste the URL, save.
4. In a chat, check the connector is enabled in the tools menu - then just ask.

It works with any MCP-compatible client, not just Claude.

When to use the built-in chat instead

We’d rather be straight with you about the tradeoffs than pretend there are none:

• Heavy number-crunching is better in the app. Our built-in chat stores large query results in a per-conversation database and lets the AI run real SQL over them - exact sums, joins across datasets, month-over-month math. Over MCP, your AI reads the raw results directly, which is fine for everyday questions but less precise when the analysis spans thousands of rows.
• Context comes built-in only in the app. The in-app chat automatically knows things like your CRM’s custom fields and your team’s saved memories. Over MCP, your AI discovers these through tools when it needs them - it works, it’s just a step less seamless.
• Reports live in the app. Scheduled reports, report history, and sharing results with your team all run through AI Data Stream itself.

Our take: the MCP connector is the perfect ad-hoc channel - quick questions, exploring your data from wherever you already work. For deep monthly reports and anything you want to keep or share, the built-in chat and skills are still the sharper tool.

Available now

External AI access is available on Pro and Business plans - team owners switch it on in Team Settings, and every member gets their personal URL from Settings → Connected AI. The setup guide walks through every step.

As always: your data stays yours, your AI stays yours, and we don’t sit in the middle charging tolls.