Your Data Stack Just Became a Conversation

You know the drill.

Open Shopify. Export orders. Open HubSpot. Export deals. Open a spreadsheet. Paste both in. Align the date columns. Realize one uses ISO dates and the other uses MM/DD/YYYY. Fix that. Build a VLOOKUP. Break the VLOOKUP. Rebuild the VLOOKUP.

Forty-five minutes later, you have a table. You still don't have an answer.

You've been a very patient human ETL pipeline. You can stop now.

What just shipped

Anna now connects directly to the tools you already use. Twenty-six integrations are live today, with another twenty-three shipping over the coming weeks. You authenticate once with OAuth — no API keys, no config files, no YAML — and Anna can pull data from your tools whenever you ask.

The point isn't that we added a list of logos. The point is that your data sources are no longer files you manage. They're live connections Anna can reach into at any time.

Where Anna connects

Here's what's live now — and what's coming next.

Live today:

• E-commerce — Shopify, Gumroad
• CRM & Support — HubSpot, Salesforce, Intercom, Zendesk, Gorgias, Attio
• Payments & Finance — Stripe, QuickBooks, Square, FreshBooks
• Email Marketing — Mailchimp, Kit (ConvertKit), Omnisend
• Surveys — Typeform
• Productivity — Google Sheets, Excel, Airtable, Notion, Monday.com, ClickUp
• Analytics — Google Analytics, Google Search Console
• Social — YouTube, Facebook Pages

Coming soon:

• Advertising — Google Ads, Meta Ads, LinkedIn Ads, TikTok Ads, Reddit Ads, Snapchat Ads
• Data Warehouses — BigQuery, Snowflake
• And more — Klaviyo, Mixpanel, Amplitude, Gong, Instagram, SurveyMonkey, WooCommerce, and others

The coming-soon list is moving fast. We're shipping new integrations every week.

The old way was fine. Until it wasn't.

Exporting a CSV from one tool and uploading it works. Genuinely. For a single dataset with a single question, it's fast and effective. Anna has always been good at that.

But here's where it breaks down.

The interesting questions — the ones that actually change what you do next — almost never live inside a single tool. They live in the space between tools.

"Which deals in HubSpot actually convert to paying Stripe customers?" That question needs CRM data and payment data. You need pipeline stages, deal sources, subscription status, and churn timing. Getting all of that into one place used to mean a data warehouse, a SQL analyst, and a Jira ticket with a two-week turnaround.

Or it meant you just... didn't ask the question.

What changes when data is a conversation away

The real unlock isn't saving ten minutes on exports. It's the questions you start asking that you never would have bothered with before.

Here are four that are now one prompt away.

1. Pipeline reality check

Connect: HubSpot + Stripe

Ask Anna: "Which deals convert to paying customers — and which ones actually stick past month three?"

Your CRM says a deal is "closed-won." Great. But does that customer activate? Do they pay? Do they churn in sixty days?

Anna pulls your HubSpot pipeline data and your Stripe subscription history. She matches deals to active subscriptions, calculates conversion-to-payment rates by deal source, and flags the segments where closed-won is a vanity metric.

2. Orders meet support tickets

Connect: Shopify + Intercom

Ask Anna: "Which products generate the most support tickets per order, and what are customers actually complaining about?"

Your best-selling product might also be your most expensive to support. You wouldn't know that from Shopify alone, and you wouldn't know it from Intercom alone. You need both.

Anna pulls your order data and your conversation history, matches on customer email, and runs the cross-reference. She finds the products with disproportionate support load and clusters the ticket themes so you can see whether it's a sizing issue, a shipping issue, or a documentation issue.

That's the difference between "support volume is up" and "support volume is up because product X has a confusing return policy."

3. Survey responses meet real behavior

Connect: Typeform + Google Analytics

Ask Anna: "Do customers who report high satisfaction in our survey actually behave differently on site?"

People lie on surveys. Not maliciously — they just answer how they think they should. The real signal is whether their stated satisfaction shows up in their behavior.

Anna pulls your Typeform responses and your Google Analytics session data. She segments users by satisfaction score and compares session duration, page depth, return frequency, and conversion rate across cohorts.

Sometimes the happiest customers on paper are the least engaged in practice. That's worth knowing before you redesign your onboarding based on NPS alone.

4. Revenue meets email engagement

Connect: Stripe + Mailchimp

Ask Anna: "Which email campaigns are driving the highest-value subscribers?"

Open rates are vanity. Click rates are better. But the real question is which campaigns correlate with customers who actually pay — and keep paying.

Anna pulls your Mailchimp campaign data and your Stripe subscription history. She maps campaign engagement to revenue, churn rate, and lifetime value by cohort. You find out which emails attract window-shoppers and which ones attract the customers who stick around.

You stop optimizing for opens. You start optimizing for revenue.

What's coming next

Advertising integrations — Google Ads, Meta Ads, LinkedIn Ads, TikTok Ads — are shipping soon. So are data warehouse connections for BigQuery and Snowflake, plus Klaviyo, Gong, Mixpanel, and more.

When those land, the cross-tool scenarios get even more interesting. Ad spend meets customer LTV. Sales call analytics meets pipeline data. Warehouse queries meet live source data.

We'll write about each as they ship. For now, the integrations directory shows what's live and what's next.

What didn't change

CSV upload still works. You can still start with a blank canvas and paste in a spreadsheet. You can still drag in a file from your desktop and start asking questions thirty seconds later.

Integrations are additive. They're a new set of starting points, not a replacement for the ones that already exist.

If you have a one-off dataset from a client, a research export, or a spreadsheet you built by hand — upload it. That workflow isn't going anywhere.

But if you're connecting to the same three tools every Monday morning, pulling the same exports, cleaning the same columns — there's now a faster path.

One click, then a question

Every integration works the same way. Click connect. Authenticate with your existing login. Done.

No API keys to hunt down in a settings page. No webhook URLs to configure. No JSON config files. OAuth handles the authentication, and Anna handles the rest.

Once a source is connected, it stays connected. Anna can pull fresh data from it any time you ask — today, next week, next quarter. The data is always current because Anna reaches into the source at query time, not at export time.

That distinction matters more than it sounds. You're not working with a snapshot anymore. You're working with the live system.

The questions get bigger

Here's the thing about friction. When every question requires twenty minutes of setup, you subconsciously filter your questions. You only ask the ones that justify the effort. The small, curious, might-be-interesting questions never get asked.

Remove the setup, and the questions change.

Instead of "what were last month's sales?" you ask "what were last month's sales, broken down by acquisition channel, compared to the same period last year, adjusted for the pricing change we made in February?" Because that question now costs the same amount of effort as the simple one: a sentence.

That's the shift. Not faster answers to the same questions. Better questions, asked more often, by more people on your team.

Go connect something

The integrations directory is live. Browse by category, see what's available, connect the tools you already use.

Then ask Anna a question you've been putting off — the one that spans two tools and used to feel like too much work. It probably takes about thirty seconds now.