Lakehouse

Big Tech benefits from your data. Shouldn't you? Lakehouse transforms your confusing social media data exports into beautiful insights—all processed locally for privacy.

Comment

Inspiration

Travis worked as a social media manager for very large accounts and always wanted to transform the data to enable various business applications. When looking further, he realized that this problem extends to every human and a platform to support solving this problem needed to be built.

What it does

Lakehouse is an ETL (Extract, Transform, Load) platform specifically designed for personal data exports, initially targeting big tech user data exports from platforms like Instagram, Facebook, and Google Photos. It transforms raw, often unwieldy data exports into intuitive, interactive formats that users can actually explore and utilize. Before AI can use your data, it needs to be ETL'd. So we help regular people to get ready for AI.

How we built it

We began by thoroughly examining the structure and format of data exports from major platforms like Instagram. Through this analysis, we discovered that while users technically have access to their data, it's typically delivered in formats that are difficult to navigate or extract meaningful insights from.

We leveraged Bolt to rapidly build the product interface, creating multiple visualization components including:

  • A Google Photos-style gallery for image browsing
  • A Facebook-like timeline for chronological content review
  • A Snapchat-inspired map feature for location-based data
  • A hashtag visualizer for content analysis
  • Integration with a ChatGPT-like interface for conversational data exploration

For the backend parsing logic, we utilized Claude to build robust data processing capabilities that could handle the complex structure of Instagram data exports.

Challenges we ran into

While Bolt proved excellent for rapid UI development and iteration, we encountered limitations when trying to build complex parsing logic solely through prompt inputs. This led us to switch to Claude for the data processing components, which provided more reliable and sophisticated parsing capabilities.

Additionally, we initially struggled with collaborative development on the Bolt.new platform, as the workflow for multiple developers wasn't immediately intuitive.

Accomplishments that we're proud of

Our biggest achievement was successfully transforming raw Instagram data exports into a beautiful, Google Photos-style interface that made browsing personal photo collections genuinely enjoyable and intuitive. Seeing years of Instagram memories presented in such an accessible format validated our core vision.

What we learned

The barrier to entry for building a functional MVP has dramatically decreased with modern AI-powered development tools. However, we're curious whether the transition from prototype to production-ready software will maintain this same level of simplicity and speed.

What's next for Lakehouse

Our immediate priorities include:

  • Conducting user research to identify the most valuable ETL transformations and data sources
  • Validating market demand and willingness to pay for personal data transformation services
  • Expanding support for additional platforms and data export formats
  • Refining the user experience based on early adopter feedback

Built With

Share this project:

Updates