ClickhouseAnnouncement
The observability UI as we know it is dead. So what comes next?
Summary
ClickHouse argues that traditional observability UIs cannot satisfy every team’s unique workflows, leading many large companies to build their own. To address this, they introduce ClickHouse Observability Apps, which let users generate custom investigation experiences by describing an application in natural language. An agent builds the app with custom views, navigation, and live data, which users can refine through conversation and share with their team. The approach relies on SQL for flexible analysis, allowing queries that go beyond fixed endpoints, such as calculating processor baselines or linking payment attempts to orders. The feature is in private preview, with ongoing work on evaluation alignment, prompt injection robustness, and reusable components.
Context
The post argues that traditional observability UIs cannot satisfy every team’s unique workflows, leading many large companies to build their own. ClickHouse aims to solve this tension by letting users generate custom observability applications via AI, rather than forcing them to adapt to a fixed vendor UI.
Approach / What changed
ClickHouse Observability Apps let teams describe an application in natural language; an agent builds it inside ClickHouse Observability with custom views, navigation, and live data. Users refine the app through conversation, adapt it as questions change, and share it with their team. The apps reuse ClickHouse’s investigation APIs and core components, and they leverage SQL for flexible analysis.
Takeaways
ClickHouse Observability Apps let users generate custom investigation experiences by describing an application in natural language, which an agent builds with live data and custom views.
The approach uses SQL for flexible analysis, enabling queries that go beyond fixed endpoints, such as calculating processor baselines or linking payment attempts to orders.
The feature is in private preview, with ongoing work on evaluation alignment, prompt injection robustness, and reusable components to ensure generated apps answer the intended questions.
FAQ
How do Observability Apps work?
Users describe the application they need in natural language, and an agent builds it inside ClickHouse Observability with its own views and navigation over live data. They can then refine it through conversation, adapt it as their questions change, and share it with their team.
How does the system handle prompt injection?
The agent treats query results, existing app code, previous users' prompts, and error messages as untrusted input to reduce prompt injection risk. RBAC and sandboxing provide enforced boundaries, but the feature remains in private preview and the risk is not considered solved.
What is the main challenge with AI-generated analysis?
A query can run successfully and still answer the wrong question. Counting failed payment attempts is not the same as counting customers who couldn't pay. The team is improving evaluations to measure alignment and using results to improve the agent's context, instructions, and tools.
Source and freshness
- Source
- clickhouse.com
- Published
- Added to Yomu
Summaries are generated from the original by a language model. Check the original before relying on it.
Keep reading
Related reading
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
The best engineering reads, once a week
One useful email. Unsubscribe anytime.