Going further
Where to go once the first pipeline runs: the next thing to learn based on what you actually want to do, the three ways to drive Keboola from an agent, and where to get help.
The pipeline runs and keeps running. That is the whole shape of working in Keboola; everything else is a variation on it.
Where you are: you have finished the guide, or you already know the part you need. This page routes you by what you want to do next, not by what comes next in a sequence.
The side trips
Section titled “The side trips”The guide’s linear path is the shortest one that produces something real. These are the branches off it, each worth reading for a specific reason:
- Send a table to Google Sheets is delivery: getting a Storage table back out to where people read it.
- Load from Google Sheets and from a database are sources that need credentials rather than a public URL.
- Use a workspace is where you develop and test SQL against a copy of your data before committing it to a transformation.
- Ad-hoc data analysis explores data in a Python or R notebook instead of building a pipeline.
- Development branches change a running project safely: make the change, review the diff, then merge.
Pick your next thing
Section titled “Pick your next thing”“My real data is not a CSV file on a public URL.” Browse the data source connectors: databases, APIs, cloud storage, ad platforms, CRMs. They configure the same way the HTTP connector did, and drop into a flow the same way. Two worked examples sit in this tier: Google Sheets, a source that needs an authorized account rather than a public URL, and a database, which you can walk against Keboola’s own sample Snowflake database without credentials of your own.
“My transformation needs to be more than one query.” Transformations covers SQL, Python, R and dbt, code blocks and phases, shared code, and variables. To develop against a copy of your data interactively, use a workspace.
“I need to send data somewhere specific.” The data destination connectors cover databases, BI tools, and storage. The Google Sheets one from the side trip is the simplest of the family.
“I do not want to break production while I experiment.” Development branches let you change configurations, run them, and review a diff before merging anything into production.
“I want to explore data rather than build a pipeline.” Do ad-hoc analysis in a Python or R workspace, or query Storage directly from a SQL workspace.
“Other people need this data.” Publish it to the Data Catalog so other projects in your organization can link it read-only instead of copying it.
“It needs to be reliable.” Set up notifications, read Jobs when something fails, and follow the best practices cheat sheet for naming and structure before the project grows.
“The connector I need does not exist.” Build it. The Component Quick Start walks writing and packaging one, and joining or creating a vendor comes with access to a development Keboola project to test it in. Making it available to everyone else is a separate, approval-gated step: publishing.
Drive Keboola with an agent
Section titled “Drive Keboola with an agent”Everything you just did by clicking can be done by an AI assistant instead. The platform exposes the same operations three ways, and which one you pick depends on where you work.
- Kai is the assistant built into the project, the one you have been using through this guide’s Prompt tabs. It already knows your configurations, table schemas, and job logs, so it is the fastest way to ask “why did this fail?” or “write me the transformation that joins these tables.” No setup. Use cases covers what it does beyond the arc: dashboards, project documentation, table descriptions, cohort analysis.
- MCP server connects Keboola to an AI assistant or IDE that speaks the Model Context Protocol (Claude, Cursor, and others). Use this when you want your coding agent to read and change the project while you work on code.
- kbagent CLI is Keboola from the terminal: projects, configurations, jobs, storage, and flows, scriptable, and usable by a coding agent that has a shell. Start with the CLI quickstart.
If you are documenting or automating your own work on top of Keboola, the API reference is the layer underneath all three.
When you get stuck
Section titled “When you get stuck”- Support: the support form inside your project reaches us with the project context attached.
- keboolastatus.com: platform status, worth subscribing to.
- changelog.keboola.com: what shipped recently.