Pixeltable announced the launch of its open-source AI data infrastructure, backed by a $5.5 million seed round led by The General Partnership, with participation from Exceptional Capital, South Park Commons, Liquid 2, Serena Data ventures, and notable industry veterans like Michael Stoppelman (Angel Investor & Former SVP of Engineering at Yelp), Wes McKinney (Principal Architect, Posit), Bill Hsieh (Bridge Street Advisors), and Steven Mih (CEO Across AI, former CEO of Aviatrix, Alluxio, and Ahana).
As AI applications become multimodal and complex, teams spend more time managing infrastructure than innovating. Pixeltable solves this by providing a unified declarative interface that handles multimodal data workloads, incremental updates, and lineage tracking, reducing infrastructure code by up to 80% and compute waste by 50%.
Pixeltable’s key innovations include:
1.) Unified multimodal interface – Handle video, images, audio, and text with structured and unstructured data living side by side through a consistent, intuitive table API
2.) Automatic incremental updates – Only process new data, eliminating redundant computation
3.) Combined lineage and versioning – Track transformations from data to model inferences in one place
4.) Development-to-production mirror – The same code works in both environments without rewrites
5.) Flexible integration and extensibility – Use built-in and custom Python functions (UDFs) while building tables through any standard frameworks and formats
For example, complete video, audio, and frame processing workloads can be built in just a few lines without sacrificing algorithmic flexibility.
The power of Pixeltable’s approach is highlighted through PixelBot (code), a context-aware Discord chatbot that showcases how Pixeltable solves current challenges in AI development like maintaining embedding indices and providing data lineage and versioning from raw data to LLM outputs.
Beyond computer vision, Pixeltable has been seeing strong adoption in generative AI applications, especially for Retrieval-Augmented Generation (RAG). And the unified approach to data management streamlines RAG workflows by combining document storage, embedding computation, and incremental indexing.
Early adopters report significant improvements:
— Reduction in infrastructure code
— Decrease in computing costs through incremental processing
— Reduction in development time
— Zero infrastructure management overhead
The company was founded by:
— Marcel Kornacker, founder of Apache Impala and co-founder of Apache Parquet, ex-Cloudera & Google
— Aaron Siegel, former Head of Data Platform at Airbnb and Twitter, and Head of Data at Chainlink
— Pierre Brunelle, former CEO of Noteable (acquired by Confluent), ex-Amazon
The seed funding will accelerate Pixeltable’s development, focusing on expanding core infrastructure capabilities, building collaboration features focused on multimodal data management, and developing Pixeltable Cloud.
KEY QUOTES:
“Just as relational databases revolutionized web development, Pixeltable is transforming AI application development. Our platform removes the need for in-house scripting of multimodal data management and orchestration, and reduces compute costs and maintenance overhead, enabling teams to deploy production AI applications in days instead of months.”
– Marcel Kornacker, CTO and co-founder of Pixeltable
“Most AI teams face a difficult choice today. Either they spend months building complex infrastructure for production-ready AI applications, or they use high-level frameworks that prevent critical use cases and limit their ability to innovate. Pixeltable eliminates this tradeoff. With our declarative approach, developers can build production-grade AI applications with infinite memory and real-time context awareness in less than 100 lines of code, while maintaining full control over their custom application logics. Teams no longer have to choose between ease of development and control over their application logic.”
– Pierre Brunelle, CEO and co-founder of Pixeltable
“Pixeltable has transformed our computer vision workflow. Before Pixeltable, our engineers spent 80% of their time on data plumbing. Now, they can focus on what truly matters — building better models and delivering value to our customers. With Pixeltable, we’ve reduced our infrastructure code by 90%, cut compute costs by 70%, and dramatically accelerated model iteration cycles. Its seamless integration into existing vision workflows means our team can rapidly experiment and iterate, delivering high-impact solutions faster than ever.”
– Adil Mohammad, Founding Engineer at Obvio and formerly Senior Deep Learning Engineer at Nvidia
“Pixeltable has transformed how our ML engineers spend their time. We have a highly multimodal use case, and Pixeltable gives us full visibility into the input data, models, and incremental steps of our system’s pipeline.”
– Denise Kutnick, CEO of Variata, formerly Director of MLSys Product at OctoAI
“At Airbnb, I witnessed firsthand how ML teams struggle with data. Teams often spent more time on data plumbing than on developing their actual application logic. The recent proliferation of generative AI tools and increasing multimodality of ML workloads has only made the situation worse.“
– Aaron Siegel, co-founder and Chief AI Officer at Pixeltable