09 Wrap-up

Starting point

You are on build-workshop-v1 — the complete app you have been configuring all session. If you checked out a fault branch in module 07, use that module's stash-and-switch reset before continuing. Budget about 10 minutes.

Why

You leave with a running prototype and this repo to show your team. Three hours of building goes by fast, so this module makes what you did explicit: what you can demo, what keeps running after today, and the one change that repoints it at your own systems.

Goal

Be able to name each piece you built and know the one thing you would change to point it at your own data.

Step 1 — Review what you built

You took the NYC-taxi app and:

• Moved its historical analytics onto ClickHouse Cloud, seeded from object storage (module 01).
• Added real-time CDC from managed Postgres via a ClickPipe (module 03).
• Explored the data conversationally with ClickHouse Agents (module 04).
• Made the app observable with ClickStack and HyperDX (module 05).
• Had your agent build an SRE dashboard and alert over that telemetry (module 06).
• Diagnosed and fixed a live incident with the AI SRE (module 07).
• Wired an in-app AI chat traced end to end in Langfuse (module 08).

Step 2 — Take it home

Your ClickHouse Cloud resources persist for the trial: the ClickHouse service, managed Postgres, ClickPipe, ClickHouse Agent, and ClickStack artifacts. Langfuse traces stay in your Langfuse Cloud project. The app, trip writer, and telemetry collector are local, so restart Docker when you return. Save .env.workshop privately. That is enough to walk your team through five things live:

• a real-time operational dashboard on ClickHouse Cloud (modules 01-03),
• a change-data-capture pipeline streaming from managed Postgres (module 03),
• conversational BI over your data with ClickHouse Agents (module 04),
• an AI-built SRE dashboard and alert over your telemetry (modules 05-06),
• and an in-app AI chat traced end to end in Langfuse (module 08).

Step 3 — Extend it to your own data

db/ and .env.workshop

The shortest path to value is to repoint the pipeline at your own systems:

- Point a Postgres CDC ClickPipe at YOUR operational database instead of the workshop's managed instance.
- Adjust the schema and the back end's queries to your tables.
- Send YOUR app's OpenTelemetry to ClickStack and rebuild the SRE dashboard on it.

How to verify you are done

• You can describe the end-to-end architecture without looking.
• You know which single config change points CDC at your own database.

Wrap-up

You built a real-time, observable, AI-assisted analytics application on ClickHouse Cloud in one sitting, with an agent doing most of the typing. That is the take-home: this is repeatable on your own stack.

You're done — what next

Your managed cloud resources persist for the trial; restart the local app when you want to use them. When you are ready to take it further, pick any of these:

• Bring your own data. Repoint the Postgres CDC ClickPipe at your operational database and adjust the schema and queries to your tables (module 03 shows the shape).
• Go deeper on ClickHouse Agents. Point an agent at your own datasets at ai.clickhouse.cloud and explore conversational BI beyond the taxi data.
• The repo is yours. Keep this repository and your saved .env.workshop; rerun any module, or hand it to a teammate — it runs fully self-paced.

If something is broken

• Traces or data not appearing? Re-check the relevant You should see checkpoint in that module — it names the exact expected output.
• Setup or credential issues? Troubleshooting lists every failure seen in testing with its fix.
• Want a clean baseline? git switch build-workshop-v1 returns you to the complete app.

End state

You have finished the workshop. Explore the ClickHouse docs and revisit any module to go deeper.