California-based ApertureData Inc., the developer of a purpose-built database for artificial intelligence multimodal large language models, today announced it has raised $8.25 million.
The seed round was led by TQ Ventures with participation from Westwave Capital and Interwoven Ventures.
As more AI-driven applications expand beyond pure text generation and answers, many have become multimodal, which means they can process not just text but images, audio, video and sometimes complex unstructured data. Although these data types can be processed individually with ease, they often live in different databases throughout the enterprise information technology stack and are accessed through varied pipelines. That makes approaches that use them difficult and tedious to build.
ApertureData’s solution ApertureDB is one database to handle all these datasets in one place. It works with text, images and videos, and can manage unstructured blobs of arbitrary size and complexity. It can store the metadata, unstructured data, and support data all in one place so that it can be fed directly into the AI model without the need for it to be pulled from multiple sources and pipelines for processing.
“The increasing adoption of multimodal data in powering advanced AI experiences, including multimodal chatbots and computer vision systems, has created a significant market opportunity,” said ApertureData Chief Executive Vishakha Gupta.
Numerous companies have rolled out or enhanced multimodal AI models, including Google LLC with Gemini Pro and OpenAI with GPT-4o.
Multimodal data fits into many industries, for example, healthcare where doctors need to review CT scans or X-rays, as well as blood tests to diagnose patients and prepare treatment plans. They also take audio notes using dictation, get data from EKG machines and images of written forms.
These various formats are unstructured and the data cannot be easily represented in spreadsheets or relational databases without transformation, and they’re not simple to search through. However, they underly the fundamental relationships of human interaction with the world.
By streamlining the separate processes of storing, transforming and preparing the data into one place, ApertureData says, it can reduce the time data scientists and AI engineers need to spend on infrastructure. That can accelerate development time from months to days. According to internal metrics, the company said that ApertureDB is 35 times faster than existing siloed solutions at preparing multimodal datasets and two to four times faster than other open-source vector databases.
Alongside the funding, the company announced the launch of ApertureDB Cloud, a fully managed cloud-based unified solution that allows businesses to centralize all these datasets in one place. The service is available for AWS and Google Cloud, with Microsoft Azure available for custom configurations.
The multimodal AI market is growing rapidly, according to data from market insights firm MarketsAndMarkets and is projected to grow to $4.5 billion by 2028 from $1.0 billion in 2023. Some key influences include advancements in AI and machine learning, demand for improved user experience and the increased integration of AI into consumer and enterprise applications such as its use in healthcare and other industries.
The company said it would use the funds to scale its current production deployments and enhance user experience by improving documentation and sandbox environments. It also intends to broaden its ecosystem integrations and expand its sales and marketing efforts to follow the wave of multimodal AI-driven apps being built.