GCP – Winners of the 2024 Looker Hackathon (Vertex AI Edition)
The 2024 Looker Hackathon (Vertex AI Edition) was a special one, as we invited our developer and data communities to collaborate, learn, and inspire each other, with a focus on how to use Vertex AI, Google Cloud’s AI development platform, with Looker and Looker Studio.
More than 160 participants from 34 countries joined, hacked away for 48 hours, and created 20 AI-driven applications, tools, and data experiences. The hacks demonstrate how developer features like Looker Studio Community Visualizations, LookML, and Looker API, in combination with Google Cloud services like Cloud Run functions, BigQuery and Vertex AI’s Gemini model, enable the developer community to build powerful, useful — and sometimes entertaining — tools and data experiences.
We are pleased to announce the winners of Hackathon and hope you take value in learning from the other finalists from this special event. In all possible cases, we have included links to code repositories or examples to enable you to reproduce these hacks.
Best Hack winners
tl;dd (Too Long; Didn’t Dashboard) – Best Hack
By Dylan A, Josh T
If you don’t think you have the time to go through the scheduled Looker dashboards emailed to you daily, this solution replaces scheduled emails of dashboard PDFs with Gemini-generated dashboard summaries.
The custom-built Looker extension lets you configure which dashboards you want to summarize and who you want to send the summaries to. You can also send additional custom instructions and context for each dashboard to Gemini to tailor the summary’s format and enhance the summary’s accuracy, and configure the extension to include prior dashboard runs and summaries so that Gemini can infer what has changed between runs and highlight the relevant information.
This hack showcases a great way to extend the functionality of Looker with the Looker extension framework and Vertex AI’s Gemini model. Check out the GitHub repository and try out this useful hack for yourself.
Looker Dashboard Review Tool – Nearly Best Hack
By Mohammed N, Paritosh M, Prashasti N
As your data organization creates and iterates on an increasing number of dashboards, how do you scalably review and enforce quality standards across all of them? The Looker Dashboard Review Tool hack can help streamline the reviews of your Looker dashboards.
Built as a custom Looker extension, the hack uses a Cloud Run function to call Gemini with a structured prompt that contains the dashboard’s LookML. The extension then displays feedback on the dashboard’s tile titles, filters, possible missing metrics, and other features. Check out the GitHub repository for more details.
Gemini Insight viz – Nearly Best Hack
By Jeremy J, Soufiane K
With this hack, you can try out Gemini’s summarization and forecasting capabilities in your Looker Explores and dashboards. The Gemini Insights viz is a great example of a Looker custom visualization that prompts Gemini with the dimensions, measures, and metadata of a Looker Explore and displays a summary or predictive insights for the data. The visualization replaces a visual representation of your data with a human-readable natural language representation.
As a Looker custom visualization, the hack can replace the visualization tiles in your Looker dashboard with Gemini’s generated summaries and forecasting. Check out the GitHub repository for more details and to try it out yourself.
Finalists
LookML code generation for visual custom drilling
By Daniela L, Peter N, Nicolas R, Joe H
If you ever wanted to drill into a measure or dimension in your Looker Explore, but found there was nothing to drill into, check out this hack that generates a visualization for drilling into Looker Explores. This hack consists of a Python Colab notebook that uses the Looker API to extract parameters from a Look and automatically generates a LookML measure that contains a visualization for drilling.
The newly generated LookML measure can be added directly to your Look to enable visualized drilling. Check out the GitHub repository for more details.
Where’s that dashboard?
By Elena G, Delphine P, Sarah T, Mary F
If you need a way to find a specific Looker dashboard among all your organization’s dashboards, this hack takes your natural language English query and uses Vertex AI’s Gemini model to return dashboards you may be interested in.
Behind the scenes, the hack accepts user input through the “Where’s that dashboard?” dashboard filter. Then the hack uses BQML to prompt Vertex AI’s Gemini model to return the most similar dashboard title from a catalog of dashboard titles fetched from the Looker API. Check out the GitHub repository for more details.
AI hackathon inspiration
This year we decided to host an extra special edition hackathon, and our developer community once again demonstrated how talented, creative, and collaborative they are. We hope these AI-focused hackathon projects inspire you to discover ways to integrate Vertex AI with your Looker and Looker Studio workflows. Tap into our linked documentation and code in this post to get started, and we will see you at our annual hackathon at the end of this year.
Read More for the details.