BeatpulseLabs raises $1.8M pre-seed to scale AI training data

BeatpulseLabs, a London-based AI data company transforming expert human judgment into high-fidelity training datasets for advanced multimodal AI models, has raised $1.8 million in pre-seed funding. The round was co-led by Araya Ventures and Lighthouse Ventures, with participation from Alumni Ventures and Avalancha Ventures.

The funding announcement comes as BeatpulseLabs reports 10x revenue growth during the first half of 2026, reflecting increasing enterprise demand for high-quality, purpose-built AI training data.

As enterprise adoption of multimodal AI accelerates, companies are facing a growing challenge: while access to raw data is abundant, creating datasets that accurately capture human expertise, context, and decision-making remains a significant bottleneck. BeatpulseLabs is addressing this gap by helping organisations transform domain-specific knowledge into production-ready training data.

Founded by South African Jason Rieff and Bulgarian Nikolay Vitanov, BeatpulseLabs was created to address a fundamental limitation in artificial intelligence. Many multimodal models continue to be trained on poorly annotated or generic datasets, reducing their ability to perform reliably in real-world environments where context and nuanced human judgment matter.

According to Vitanov, enterprise AI often encounters challenges when moving from controlled testing environments into real-world operations. He said BeatpulseLabs addresses this by creating training data that reflects how individual businesses actually function:

We proved this approach in some of the most demanding multimodal domains such as music, video and speech. The same logic applies anywhere the margin for error is low, from robotics to knowledge work. Using generic training data is like letting a confident stranger make decisions for your business. We do not recommend it.

BeatpulseLabs offers two integrated services: dataset preparation and dataset provision. The company transforms existing multimedia content libraries into enterprise-grade AI training datasets by cleaning, structuring, labelling, validating, enriching, and formatting raw speech, music, and video assets for machine learning applications. It also provides ready-made and custom rights-cleared datasets for organisations seeking high-quality training data without relying solely on their own content archives.

These datasets are designed to support model training, fine-tuning, reinforcement learning, and evaluation, enabling AI systems to operate with greater accuracy, context awareness, and reliability.

Rieff emphasised that the capabilities of AI systems are largely determined by the quality of their training data, noting that much of the data currently used is broad, inconsistently organised, and inadequately annotated for enterprise use cases.

We are building the missing data layer by transforming raw multimedia content into structured, annotated, model-ready datasets that help AI systems understand context, not just patterns. The traditional approach of applying broad labels to large volumes of content is no longer sufficient for the next generation of AI.

The funding will support BeatpulseLabs as it expands its platform and customer base amid growing demand for high-quality, domain-specific AI training data.