Less Talk, More Product: What’s New in Sifflet This Month (May 2025)

Hi everyone!
Two weeks ago, we launched our Agentic AI system, a big step toward the future of autonomous data observability. It’s a bold bet on where things are headed: systems that reason, remember, and act on your behalf. And we’re excited about it. Really.

But we also know this: even the smartest agents won’t help if the basics still feel like a grind.

So while the AI buzz keeps building, we’ve also stayed focused on another part of the mission: shipping the kinds of improvements that make your workday smoother now. The kind of stuff that helps you trust your monitors more, set up integrations faster, and close your laptop earlier on a Friday afternoon.

In this update, you’ll find things that are not necessarily flashy, but useful for sure.

And we’re proud to be building both: the long-term vision and the week-to-week fixes that make observability feel less like firefighting and more like flow.

Let’s get into it.

‍

‍Meet the MCP Server: Observability Meets Developer Flow

We’re introducing the MCP (Model Context Protocol) Server, a new way to interact with Sifflet directly from your development environment. We told you everything about it here but think of it as your gateway to programmatic observability: you can query assets, review incidents, generate monitors as code, and even trace lineage, all without leaving your IDE or notebook.

Why it helps:

If you're a data engineer or analytics developer, this means no more switching between tools or guessing context. You can stay where you’re most productive and still get full visibility into the health and structure of your data.

‍

Custom Metadata: Your Language, Your Logic

You can now add Custom Metadata to any asset, monitor, or domain in Sifflet, in the form of flexible key-value pairs. And if you watched this interview of Salma Bakouk, our CEO, with George Firican, you should already know we're big on metadata.

Ownership, SLAs, lifecycle stage, business unit… whatever makes your data easier to organize, route, or find, you can now define it in your own terms.

Why it helps:

We know every org structures data differently. This feature puts you in control, whether you're trying to filter dashboards by team or surface high-priority monitors by SLA level. It also lays the groundwork for smarter alerting and routing down the line.

‍

Build Joins Right in the Monitor UI

You asked, we delivered: you can now define joins across datasets directly in the monitor setup interface.

Previously, you had to manually write JOIN logic or pre-materialize your data. Now, just point and click: combine tables, add conditions, and get the context you need to catch cross-table inconsistencies.

Why it helps:

Let’s face it, most data quality issues don’t live in isolation. Whether you’re comparing orders to transactions, or leads to accounts, joins let you monitor data relationships, not just rows in a single table.

‍

Distribution Change Monitor: Smarter, Sharper Insights

We’ve upgraded how our Distribution Change monitor works. Instead of testing the whole distribution statistically, it now tracks how much volume shifts between specific categories (e.g. product lines, customer segments, statuses).

Why it helps:

Less noise. More relevance. You’ll get alerted when something important actually changes (like if a key segment drops to zero) not when minor statistical blips occur.

Bulk Manage Users by Domain (Finally)

We’ve added a dedicated Users section inside Domain Management, so admins can now manage access in bulk and with clarity.

This was a big request from teams working across multiple domains, and we’re glad to finally simplify it.

Why it helps:

No more one-by-one edits or guessing who has access to what. You get a centralized view of user roles for each domain, and changes take seconds.

‍

Easier Setup & Troubleshooting for BigQuery and Snowflake

We’ve revamped the “Test Connection” process for BigQuery and Snowflake. It now does more than just say “failed” as it tells you exactly which permission is missing and where. Oh, and speaking about Snowflake, did you check our one-pager?

Why it helps:

You don’t need to guess or ping us for logs. You’ll know upfront if your service account is missing a role or if a policy is blocking access. It saves time during onboarding and helps prevent refresh issues later on.

‍

❤️ From Us to You

Every one of these features came from real customer conversations. You can find more details about them in the doc. We’re grateful for your feedback, your honesty, and your wild use cases as they push us to keep making Sifflet better.

If you’d like a walkthrough of anything mentioned above, your CSM would love to hear from you. Or just drop us a message directly.

More to come (as always)

‍