Typedef Inc. recently managed to officially move out of Stealth Mode, having secured $5.5 million in seed funding led by early-stage investors Pear VC, along with additional participation from Tokyo Black, Verissimo Ventures, Monochrome Ventures, and several additional angel investors.
With co-founders Yoni Michael and Kostas Paradalis at the helm, two successful data infrastructure engineers who have developed themselves into highly successful serial entrepreneurs, Typedef Inc. has seen incredible success since its establishment. Indeed, having had experience in other industries, including data analytics and more, the pair are now working towards securing market share in the modern Artificial Intelligence (AI) infrastructure market by means of a new purpose-built AI data infrastructure specifically designed for modern workloads at scale.
Michael and Paradalis are building on their leadership experience in data infrastructure at industry-leading companies including Salesforce, Starburst Data and Tecton, and their experience has allowed them to set their sights on building a solution from scratch that can handle mixed AI workloads with equal efficiency and scalability. Essentially, the objective is to operationalise AI across data stacks and the enterprise as a whole.
According to Arash Afrakhteh, Partner at Pear VC, “Typedef is ushering in the new era of AI infrastructure where model training has given way to inference and where teams can build reliable, scalable, and cost-effective Large Language Model (LLM) workloads without the complexity or strain of managing infrastructure.” According to Afrakhteh, a significant portion of his belief in Typedef is the fact that its founders have experienced the problem that they’re trying solve first hand. Because they understand what they’re dealing with, they know what they need to do to solve it properly, and they also” have the added experience of running multiple data infrastructure startups to successful exits.”
AI Projects Struggle to Deliver Real Results
Generative AI (GenAI) is firmly at the centre of today’s tech landscape, drawing major investment and strategic focus. In fact, a recent CDO Insights Survey from Informatica revealed that 93% of data leaders in the US plan to increase their GenAI spending in 2025. But, despite this enthusiasm, most organisations are struggling to translate their pilot projects into tangible business outcomes.
According to the above report, a staggering 97% report challenges in proving GenAI’s value, and two-thirds admit they haven’t successfully moved even half of their AI pilots into production. The issue, often referred to as “pilot paralysis”, continues to stall progress, with some studies suggesting that up to 87% of enterprise AI projects fail to scale, which is astonishingly high.
According to co-founder Michael, the problem lies in the difficulty of operationalising AI. “Getting AI workloads into production in a consistent, reliable way is incredibly hard,” he says. “Most companies are stuck using outdated systems not designed for the demands of large language models (LLMs), inference tasks, or unstructured data.” These limitations often lead to messy, cobbled-together tech stacks that are hard to maintain and even harder to scale. Typedef was created to solve this problem with a platform purpose-built to build, deploy and scale AI workflows that are production-ready and dependable – even when working with the unpredictability of LLMs.
A Smarter Infrastructure for AI Workflows
Typedef is a modern infrastructure platform that is specifically designed to support AI pipelines at scale – but without the operational complexity. Built specifically for LLM-powered workloads like semantic search and document processing, it handles the technical overhead, including token management and context handling, behind the scenes. With a clean, modular interface and familiar APIs, Typedef allows engineers to experiment quickly and identify which pipelines deliver the most value – then, it helps to scale them into production with ease.
The platform is fully serverless, removing the need for infrastructure provisioning or manual configuration. Getting started is as simple as installing an open-source client library and connecting your data sources. Teams can build AI or agent-based pipelines with just a few lines of code – no need for finicky and fiddly integrations or fragile workarounds!’
Co-founder Kostas Pardalis explains, “AI and data teams want the same reliability from AI pipelines that they expect from traditional analytics infrastructure. They want to turn proprietary data into insights, automate decision-making, and build AI-driven systems that work at scale. Typedef makes that possible, giving teams the tools to deliver on their AI goals without drowning in operational overhead.”
Turning Unstructured Data into Business Value, At Scale and Out In the Open
A major challenge for many organisations is unlocking value from unstructured data which, according to Gartner, makes up around 80% of all enterprise information, hidden in emails, support tickets, call transcripts and documents.
Matic, an insurance-tech firm which partners with over 70 top-rated carriers, turned to Typedef to address this issue, using its platform to build semantic extraction pipelines across thousands of policy documents and customer transcripts. The result was a significant reduction in manual effort, improved compliance and a sharp drop in errors and operational costs. ‘
“Typedef helped us roll out semantic extraction pipelines in days, not months,” said Lee Maliniak, Matic’s Chief Product Officer. “We’ve reduced costs, eliminated human error and cut down our Errors and Omissions (E&O) risk.”
Committed to open innovation, Typedef has also released a large part of its underlying technology as open source under Project Langframe, available now on GitHub. By doing so, the team aims to accelerate adoption and empower developers to build production-ready AI systems with far less friction. In a space where most AI projects still fail to scale, Typedef is bridging the gap between prototype and production, helping teams realise the real business value of GenAI.
