Hi there,

Today, we're excited to launch Web Search!

Web search is massively important for analytics agents. But it also should be very scary.

Why is it so complicated for analytics agents? Because they have access to private data, which is the third leg of Simon Willison's Lethal Trifecta. It's extremely important that web search is secure for an analytics agent who has access to sensitive company data.

What if the agent sees some malicious context from the web search and takes an unexpected action to send that data out of your system? Some of your most precious and sensitive data could be exfiltrated by a hacker just like that. In seconds.

That's why we've designed a custom web-search implementation natively for data agents. It's not possible for our agent to exfiltrate data because we carefully craft its tools so that capability doesn’t exist.

With secure web search, our customer use cases are incredible:

• One retail customer had a competitor shut down several stores. They used web search to find those store locations and launch targeted promos in the area, tuned with their internal marketing data.
• One manufacturer customer tracked how they meet their SLA commitments to their customers, and used web search to find the latest news that might be impacting that customer.
• A software customer cross-references product launch data with competitor activities, to understand how usage spikes and then decays after big launches.

Web search with your data is amazing (when it's secure). Here's what that looks like →

Hi there,

It's really easy for any AI interface to be a giant stream of text from the agent that's visually exhausting for the human. Honestly, we had that same problem with Zoë.

Zoë can do a ton of work, but when she does all that work, you used to see a long string of accordions where Zoë was taking different actions and building resources and iterating on the results for you.

Now we've cleaned up that experience dramatically. Whenever Zoë goes to answer a question she'll group her bodies of work together in easily understandable headings and you can just focus on the results you care about, without sacrificing the ability to dig deep into how Zoë got the answer.

I love how the new UI is looking and you can check it out here →

Hi there,

Artifacts are now in public preview!

We're excited to launch Artifacts out of beta and into a public preview. All Zenlytic customers will now have access to artifacts and be able to create incredibly rich data readouts for themselves and their teams.

Artifacts are a really deep feature set. So this is something I'm going to be diving into over the next few weeks to highlight more on what you can do with them.

To start, this is my favorite jumping off point that makes it easy for people to get started and build something amazing in just a few minutes.

Artifacts → Create Artifact → Watch the magic happen

Hi there,

We've made several updates to how you work with Zoë.

Chat history is now grouped and collapsible, making it easier to find past conversations. You can also run multiple chats at the same time — start a new conversation while another is still running.

Data Model Editor & Workspace Settings have moved to the bottom of the sidebar. Depending on your role, you may or may not see these.

Account menu (bottom left) now includes Workspace Manager and Support (formerly Help).

Everything works the same — just easier to find.

Every team works differently, and now Zoë can too.

Workspace Manager gives your admins the ability to create dedicated agents for different teams and domains across your organization. Each agent can be tailored to the workflows, terminology, and data needs of the people using it, so finance Zoë speaks finance, and ops Zoë speaks ops.

For enterprise organizations managing complex data environments, this means you can scale Zoë’s impact across the entire business, not just the teams who were first to adopt her.

Ready to set it up? Head to your Workspace Manager to create new agents.

Hi there,

We are launching an improved and more intelligent version of Zoë using what we're calling exploratory mode. This update gives Zoe the ability to answer more complex questions while still fully governing and explaining how she gets to her answers.

Here's a quick 1 min video summarizing her new capabilities

If you prefer the previous mode of interaction, you can hit X on the exploratory mode icon in the conversation to go back to the previous experience.

Let us know your feedback!

Hi there,

Last year, we launched the world’s first proactive analytics agent… and then immediately realized we had no idea what to name it. Was it a Task? A Workflow? An Automation? At one point, “automagics” was on a whiteboard. Truly dark times.

We went with 🌟Intelligent Workflows🌟

and kept building. But now it's time for a name that actually fits.

Yes, it’s a rename, but it’s also a reflection of how far the product has come. It now delivers truly proactive insights, flags issues automatically, and keeps getting smarter.

Learn more click here

Hi there,

The Dynamic Fields Admin Panel in Clarity Engine shows you exactly what concepts your users are asking Zoë about and how frequently.

When users ask about concepts that don't exist in your data model yet, they get captured here. You can review the requests, iterate on definitions with your business teams, and promote the ones that matter into your standardized model.

No more guessing what to build next. And you're never the bottleneck.

Check it out under Settings → Dynamic Fields and let us know what you think.

Hi there,

With the Clarity engine, Zoë has the ability to create new metrics on the fly.

To ensure Zoë’s answers don’t feel like a black box, when Zoë creates a new metric, she’ll give you an “Explain” button where she will explain exactly how she’s calculating the metric you asked for making it easy for you to check Zoë’s work.

This makes it even easier to trust Zoë’s answers!

LLMs give different answers to the same question by default. Ask for "Latin America sales" twice, and you might get Mexico included once and excluded the next time.

That’s why we’re launching Memories.

Memories lock in those nuanced assumptions permanently.

How it works:

1. Ask Zoë any question
2. Click "Add to memory" when you get the right answer
3. Zoë remembers your exact methodology forever

What you get:

• Same answer, every time: No more conflicting results
• Team-wide consistency: Everyone gets identical answers
• No setup required: Build your knowledge base naturally as you work