ClickHouse and Hud Launch Runtime Feedback Loop for AI-Generated Software

Software observability startup Hud and open-source database developer ClickHouse have introduced a strategic integration connecting ClickHouse's ClickStack observability suite with Hud's Runtime Code Sensor. The collaboration aims to create an automated feedback loop for code generated by artificial intelligence, allowing development teams to assess, deploy, and remediate machine-written software using real-time production data, as first reported by The Next Web.

The partnership addresses a major structural shift in modern software engineering. According to figures published by Hud, artificial intelligence currently generates or assists in writing 42 percent of all production code, a share projected to expand to 65 percent by 2027. While automated tools have significantly accelerated initial code creation, engineering teams increasingly face bottlenecks during code review, release verification, and production troubleshooting.

To solve these operational hurdles, the combined workflow pairs ClickHouse's open-source ClickStack telemetry platform with Hud's code-level tracing tools. ClickStack delivers high-level operational visibility across applications and cloud infrastructure, identifying specific endpoints and services experiencing issues. Meanwhile, Hud monitors function-level behavior and links execution data directly to the specific code commits responsible. Coding agents can connect the two environments using shared trace IDs, enabling engineers to jump from infrastructure alerts in ClickStack straight to underlying functions in Hud.

“AI is accelerating how quickly teams can generate code, but safely shipping it with high confidence requires production context,” said May Walter, Chief Technology Officer of Hud. Walter stated that bringing comprehensive production behavior across the entire software development lifecycle allows teams to gate high-risk changes before deployment, proactively verify new releases, and resolve emerging software bugs using real runtime insights.

Under the integrated framework, code modifications undergo pre-deployment risk evaluations against real-time operational metrics. Changes deemed low-risk can be merged automatically or moved rapidly through build pipelines, while higher-risk edits are held for human review. Once deployed, the system monitors software behavior, automatically executing rollbacks if regressions occur or generating automated pull requests via agentic workflows to fix underlying software issues.

“Our users already trust ClickHouse to store and query their Open Telemetry data at scale,” said Mike Shi, Head of Observability at ClickHouse. Shi noted that software engineering is shifting from passive infrastructure monitoring to actively utilizing telemetry data for daily operational choices, making combined observability stacks far more useful for organizations shipping code built alongside AI tools.

Enterprise workplace platform monday.com has signed on as an early adopter of the combined solution. “Like every modern engineering organization, a growing share of our code is now written with AI,” said Rom Kadria, Senior Software Engineer at monday.com. Kadria noted that while AI speeds up writing code, the primary challenge remains shipping safely, adding that combining ClickHouse's wide operational view with Hud's runtime introspection allows the company to triage and resolve production incidents rapidly.

Engineering organizations can deploy the system by installing the Hud software development kit (SDK) and connecting it directly to their existing ClickStack infrastructure. Hud's function-level telemetry subsequently runs alongside standard OpenTelemetry data streams, feeding operational behavior directly back into automated development workflows.