Open House observability announcements: MCP server, AI Notebooks, and ClickStack Cloud<br>Open searchOpen region selectorEnglish<br>Japanese

47.7kSign inGet started

->Scroll to top

BackBlog<br>Product<br>Copy pageCopied!More actionsView as Markdown Open this page in Markdown<br>Open in ChatGPT Ask questions about this page<br>Open in Claude Ask questions about this page<br>Open in v0 Ask questions about this page

Open House observability announcements: MCP server, AI Notebooks, and ClickStack Cloud

Mike Shi and Brandon Pereira<br>May 27, 2026 · 13 minutes read

Open House brought the ClickHouse community together for three days of workshops, technical deep dives, product announcements, demos, and conversations about what’s next for real-time data. We were glad to meet so many users, customers, and members of the observability community throughout the event.

For those who couldn’t join us in person, here’s a recap of the observability announcements we shared at Open House.

We announced three major updates across ClickStack and observability: ClickStack Cloud, AI Notebooks in beta, and a new ClickStack MCP server.

ClickStack Cloud #

The biggest announcement, and one that deserved its own blog post, was the introduction of ClickStack Cloud in private preview.

ClickStack Cloud is a fully managed, serverless observability platform built on ClickHouse. Instead of managing collectors, infrastructure sizing, scaling policies, or schema tuning directly, users simply send OpenTelemetry data to a managed endpoint and immediately start exploring logs, metrics, and traces through the ClickStack UI.

ClickStack Cloud is aimed at reducing that operational work while still keeping the performance characteristics people love about ClickHouse.

For more details, we recommend the dedicated post.

Managed ClickStack is generally available #

In addition to ClickStack Cloud entering private preview, our existing Managed ClickStack offering is now generally available.

Managed ClickStack is designed for teams that want direct operational control over their observability stack, including ingestion pipelines, compute sizing, workload isolation, schema design, and datastore tuning. Users manage their own OpenTelemetry collectors and ingestion architecture while using ClickHouse Cloud as the underlying observability datastore. For many large-scale deployments, that control is essential for optimizing performance and achieving market-leading cost efficiency.

Managed ClickStack and ClickStack Cloud are designed for different operational models.

As discussed above, ClickStack Cloud will provide a fully managed, serverless observability experience where teams send telemetry to a managed endpoint and immediately begin exploring logs, metrics, and traces without managing infrastructure directly. Conversely, Managed ClickStack is intended for organizations that want deeper control over scaling strategy, ingestion architecture, and workload optimization while still running on ClickHouse Cloud infrastructure. Together, the two offerings give teams a choice between a turn-key observability experience and a more configurable platform for operating observability at scale.

AI Notebooks in Beta #

We also announced AI Notebooks entering beta for Managed ClickStack.

Over the last year, nearly every observability platform has added some form of AI chat experience, but we increasingly felt that chat alone does not match how real incident investigations actually unfold. Production debugging is messy, and engineers jump between logs, traces, dashboards, deployments, and hypotheses. They backtrack, split into parallel investigations, and revisit earlier assumptions as new signals appear. Incidents are rarely single-threaded conversations, so we did not want the interface to force them into one.

Investigations are rarely single-threaded. SREs typically need to explore multiple branching hypotheses before reaching a resolution.

AI Notebooks are designed as a persistent investigative workspace rather than a transient chat session. Each investigation becomes a structured sequence of prompts, queries, charts, reasoning steps, and findings that remain visible and editable throughout the process.

Engineers can branch from any point in the notebook to explore alternative theories without losing previous work or context. In practice, the workflow feels more like a collaborative debugging experience.

Loading video...

We were also pretty opinionated about transparency while building this. In a production incident, engineers need to understand what the system is actually doing, especially if AI is involved in the investigation loop. Every query, chart, reasoning step, and intermediate result is visible inside the notebook. You can edit queries manually, insert your own searches, or ignore the suggested path entirely and take the investigation somewhere else. We wanted the AI to behave more like a collaborator sitting beside the engineer than a system producing...