We’re excited to announce that ClickHouse has achieved the AWS AI Competency in the Software partner track, with validated offerings in Agentic AI Tools and Agentic AI Applications.
As organizations move beyond experimentation toward production-ready autonomous systems, AWS recently expanded the AWS AI Competency (formerly the AWS Generative AI Competency) with three new agentic categories: Agentic AI Tools, Agentic AI Applications, and Agentic AI Consulting Services, to help customers identify partners with proven expertise in delivering enterprise-grade autonomous AI solutions.
Meeting the unique demands of agentic AI #
Agentic AI systems don’t just “generate”, they observe, plan, act, and learn. That introduces a data problem that looks very different from classic analytics:
-
High-volume, high-cardinality telemetry (tool calls, traces, evaluations, feedback signals, token usage, latency, errors)
-
Real-time iteration loops (prompt changes, model rollouts, A/B tests, regression checks)
-
Enterprise governance requirements (data locality, private networking, predictable cost, operational resilience)
ClickHouse is built for this kind of workload: a fast, scalable real-time analytics engine that can ingest massive event streams and return interactive query results quickly, so teams can keep humans and agents in the loop without slowing down production systems.
Earning the AWS AI Competency recognizes the work we’ve done (with customers) to bring these capabilities to production on AWS, and it reflects AWS’s bar for partner validation and demonstrated customer success in agentic categories.
Customer success across AI tools and agentic applications #
We see two consistent patterns in production AI: teams need (1) a reliable data backbone for AI tools (observability, evaluation, iteration), and (2) a low-latency analytics layer to power agentic applications that interact with real-world business data. Here are six examples that reflect those requirements:
LLM observability and evaluation at scale (Agentic AI Tools)
Platforms like Langfuse and LangChain’s LangSmith rely on ClickHouse to store and analyze the traces, metrics, and feedback that teams use to debug and improve LLM and agent behavior. As these systems scale to production volumes (including self-hosted environments), ClickHouse enables fast dashboards, drilldowns, and evaluation workflows without the ingestion and query bottlenecks that can emerge in general-purpose architectures.
Production-grade telemetry for AI infrastructure (Agentic AI Tools)
Modern AI infrastructure generates an immense amount of operational signals. Modal uses ClickHouse Cloud to power real-time observability dashboards—ingesting 1–2 million events per minute while still delivering sub-second queries. Temporal built Chronicle, an internal observability system powered by ClickHouse, to “embrace cardinality” and keep investigative queries and dashboards fast at multi-tenant scale.
AI-powered analytics experiences and domain applications (Agentic AI Applications)
On the application side, Property Finder modernized its analytics foundation on AWS with ClickHouse, achieving major performance gains and cost reductions, while also enabling AI-assisted analytics workflows, including natural language requests translated into SQL using Amazon Bedrock, and predictive modeling with Amazon SageMaker. In healthcare and life sciences, Memorial Sloan Kettering (cBioPortal) uses ClickHouse to accelerate large-scale genomic analytics, and the cBioPortal team is also developing an AI-powered chat experience built on LibreChat with Claude via Amazon Bedrock, connected to cBioPortal data through MCP servers, with queries executed against ClickHouse.
Beyond these six stories, you can also explore how teams including mpathic, SewerAI, RBC Borealis, poolside, Cognitiv, Lens, DENIC, AdMixer, and DeepL are using ClickHouse for machine learning and GenAI workloads.
Looking ahead #
Agentic systems raise the bar for data infrastructure, especially as organizations demand stronger governance, lower latency, and clearer observability across increasingly autonomous workflows. We’re continuing to invest in helping teams build and run agentic solutions on AWS, from the ClickHouse core engine and ClickHouse Cloud to the open-source Agentic Data Stack that connects natural language interfaces to large-scale analytics.
If you’re building agentic AI tools or applications on AWS and want to accelerate from prototype to production, we’d love to work with you.
