A new player is emerging from stealth with backing from Google’s AI-focused venture fund to help businesses assemble their open source AI infrastructure and curb engineering overhead. Cake is designed to weave together and secure more than 100 enterprise components, spanning data source adapters like Apache Hadoop, data ingestion systems such as Apache Kafka, data labeling tools like Label Studio, vector and graph databases such as Milvus and Neo4j, and the generative AI APIs and related tools from providers such as Anthropic, among many other categories. The name itself hints at the company’s broader aim: to take the diverse “layers” that constitute the AI stack and encode them into a production-ready, digestible format tailored for business environments.
The big picture problem: why Cake exists and how it addresses it
Cake was born out of a recognition that the modern AI stack is not a single monolithic project but a sprawling ecosystem of open source components, each with its own strengths, quirks, and evolving interfaces. For enterprises, the challenge is not merely selecting a handful of tools but orchestrating a cohesive, reliable, and secure environment where dozens or even hundreds of components can work in concert. This is the essence of the “big picture problem” that Cake seeks to solve.
In practical terms, many organizations want to harness retrieval augmented generation (RAG) or other advanced AI capabilities over vast repositories of data. When a traditional off-the-shelf product falls short—either in capability, nuance, or compliance—teams face a time-consuming, high-friction process: they must assemble multiple components, ensure they work together, authenticate and authorize access, monitor for security gaps, and maintain production-grade reliability. Cake positions itself as a solution that not only curates a curated selection of open source projects across the entire stack but also delivers a managed, production-ready layer that makes these components behave as a cohesive whole.
For large financial services firms, the promise is clear: millions of documents containing complex financial data can be subjected to sophisticated AI workflows, such as RAG, to improve the quality and relevance of natural-language responses. If current offerings do not meet stringent compliance requirements or fail to scale, companies may be forced to embark on bespoke integration projects. This is precisely the kind of do-it-yourself undertakings Cake aims to simplify, taking on the heavy lifting of integration, reliability, and security so customers can focus on deriving actionable insights rather than wrestling with infrastructure.
The hospital example illustrates another axis of Cake’s appeal: secure systems for analyzing medical imagery, such as CT scans, require careful handling of sensitive data, strict privacy controls, and robust, enterprise-grade reliability. Similarly, e-commerce platforms looking to upgrade their recommendation engines face the need to blend multiple data streams, model types, and deployment pipelines in a way that preserves performance at scale, while meeting privacy and regulatory standards. Across these examples, Cake’s sweet spot emerges as helping teams move beyond basic, ready-to-use products toward a more capable, integrated AI infrastructure that remains approachable for smaller teams while delivering enterprise-grade reliability.
Cake’s founder and leadership describe the platform as a way to “make sense of the myriad open source components” that constitute today’s AI stack and to provide a bundled, managed, open source AI infrastructure that is accessible to small teams and growing organizations. Rather than building a single open source project and banking on a singular success story, Cake emphasizes assembling and curating a selection of projects across the entire stack and ensuring they can run smoothly in enterprise environments. This approach is meant to remove the friction that typically accompanies adopting cutting-edge AI tooling, especially when teams lack the headcount or deep specialist expertise required to integrate disparate components.
In practical terms, the platform is designed to enable production-grade implementations of AI workflows even when the underlying tools originate from different development communities, with attention paid to operational concerns such as monitoring, security, governance, and data integrity. The overarching narrative centers on enabling enterprises to implement advanced AI capabilities—such as sophisticated data querying, retrieval, labeling, and model interaction—without being overwhelmed by the complexity of stitching together an ever-growing landscape of open source options. This is the strategic angle that Cake aims to own: a reliable, integrated, and auditable AI infrastructure layer built on top of open source foundations.
Founding story, leadership, and funding trajectory
Cake was founded in New York in 2022 by Misha Herscu (CEO) and Skyler Thomas (CTO), both of whom bring a deep background in venture-backed technology and AI-focused engineering leadership. The company officially launched last year and began engaging with customers, though it had not been aggressively public about its activities until the formal unveiling today. Early customer relationships include Altis Labs, an AI biosciences startup focused on advanced bioinformatics, and Ping, a data intelligence insurtech. These early engagements gave Cake practical proof points in highly technical domains where robust data workflows and secure, scalable AI infrastructure are essential.
From a funding perspective, Cake has disclosed that it has raised $13 million since inception. The capital breaks down into roughly $3 million in pre-seed funding accumulated over the formative years and a recent $10 million seed round led by Google’s Gradient Ventures. The funding news underscores the company’s momentum and the strategic interest from investors who are keen on open source AI infrastructure models and enterprise-grade deployment patterns.
In interviews, Herscu emphasized that the company has not operated in stealth so much as been focused on product development and customer engagement. He noted that the team has been deeply engaged with customers, even if public disclosures were more measured until today. The fundraising narrative is augmented by Herscu’s prior experience in the AI and machine learning space, including his work founding McCoy Medical Technologies, a company focused on machine learning infrastructure for radiology that was later sold in 2017 to IT vendor TeraRecon. Following that exit, Herscu joined Primary Venture Partners in New York as an “operator in residence,” where he conducted hundreds of conversations with data science and AI executives as he shaped the next venture.
A central theme in Herscu’s storytelling is the breadth of customer discovery conducted prior to Cake’s market entry. He described performing upwards of 200 customer discovery calls to understand the systemic pain points and bottlenecks that organizations encounter when trying to harness the latest open source AI advances. The consensus from those conversations, he explained, was not that the need was to solve a single problem—such as implementing a vector database or building a data pipeline—but that there was a much larger problem: a proliferation of diverse components across a rich ecosystem, with incompatibilities, security gaps, and a lack of enterprise-grade operational models. The result is a business that capitalizes on the need to integrate, secure, and streamline a broad set of open source AI tools into a production-ready stack that can be deployed by teams of varying sizes.
In addition to Gradient Ventures as the lead seed investor, Cake’s seed round drew participation from Primary Venture Partners (the pre-seed investor), Alumni Ventures, Friends & Family Capital, Correlation Ventures, and Firestreak Ventures. The company has highlighted that the seed round completed in April and that this round signifies not only the founders’ track record but also the traction that Cake has begun to generate with real-world customers. Looking ahead, Herscu has indicated an intent to pursue additional financing, with tentative plans to raise again around the middle of 2025. He described the current state of traction as resembling what one might expect from a Series A investment, and suggested that when the company proceeds to a Series A, it could present more characteristics associated with a Series B round, reflecting maturity, growth, and expansion.
The fund emerges as a strategic partner beyond capital, given Gradient Ventures’ focus on AI-centered startups and its track record in backing early-stage teams that aim to operationalize AI in business environments. The collaboration situates Cake within a broader ecosystem of enterprise AI infrastructure companies that aim to scale the open source stack into enterprise deployments, with Gradient’s involvement signaling both validation and the potential for subsequent investor participation as Cake expands its platform and customer base.
Platform capabilities and real-world use cases
Cake’s core proposition centers on unifying and managing a large, diverse set of open source AI components into a cohesive, production-ready infrastructure. The platform’s scope spans multiple layers of the AI stack, from data ingestion and labeling to vector and graph databases and API access to generative AI models. This breadth is intentional: it acknowledges that enterprise AI workflows rely on a broad constellation of tools rather than a single solution.
A key aspect of Cake’s value proposition is the curated, bundled approach to open source projects. Instead of tethering customers to one open source project or encouraging an organization to chase after dozens of independently evolving projects, Cake offers a curated selection of open source components aligned along an end-to-end stack. The promise is to deliver a reliable, integrated experience with an emphasis on production readiness, which includes considerations such as authentication, authorization, security, scalability, and governance.
From a workflow perspective, the platform is designed to enable enterprises to deploy complex AI pipelines without the friction typical of assembling disparate components. For example, consider a large financial services organization parsing millions of documents for RAG-enabled insights. The company would want to run sophisticated retrieval-augmented processes over its data, but if no off-the-shelf product meets compliance or performance needs, it would face an extended integration project. Cake provides an alternative: a packaged, open source-based infrastructure that can be deployed in a way that aligns with the company’s security and governance standards, and that can be configured to operate efficiently at scale.
In other sectors, Cake envisions scenarios such as hospitals building secure systems to analyze CT scan images or e-commerce platforms strengthening and personalizing recommendation engines. These cases reflect a common thread: when a business requires more than basic capabilities, a robust, integrated stack can unlock higher-quality AI outcomes and better governance without forcing teams to reinvent the wheel for every project. Herscu stresses that Cake can accommodate a wide range of use cases but emphasizes that the company’s sweet spot lies in helping teams go beyond simple, off-the-shelf solutions to achieve deeper, more capable AI deployments.
From an architectural standpoint, Cake’s approach implies a focus on ease of integration and production readiness across a wide spectrum of components. The platform aims to provide a manageable, cohesive experience for teams that would otherwise spend significant time aligning authentication, authorization, data access controls, monitoring, and reliability across a heterogeneous open source environment. The result is a production-grade AI infrastructure that respects the realities of enterprise IT: security, compliance, accountability, and the ability to scale across organizational boundaries.
The leadership team has drawn analogies to established enterprise software paradigms to illustrate the ambition of Cake. Thomas, Cake’s CTO, compares the current AI open source landscape to the early Linux ecosystem, where thousands of packages existed but lacked a mature support and security model. He notes that Red Hat popularized enterprise-grade Linux by providing a reliable, supported, and secure platform on which businesses could depend. The implication is that Cake intends to perform a similar function for AI—not by owning a single core project, but by delivering a safe, enterprise-ready framework that makes it practical to deploy and operate open source AI components at scale.
According to Thomas, the enterprise adoption barrier often centers on the gap between the latest research and the needs of a corporate IT environment. Labs-produced tools may lack essential enterprise features such as robust authentication, authorization, access controls, and a clear support model, which makes them less attractive for large organizations. Cake’s strategy is to bridge this gap by offering a curated, production-ready stack that preserves the benefits of open source while delivering the operational rigor required by enterprises. This involves not only selecting the right components but also integrating them in ways that ensure consistent performance, security, and maintainability.
In addition to its on-premise approach, Cake has signaled plans to eventually offer a hosted version of its platform. The hosted option could broaden accessibility for organizations with different compliance profiles or operational preferences, though the current stance recognizes that many enterprises favor self-hosting to maintain data sovereignty and control. The leadership has suggested that cloud control of the service could be advantageous for Cake’s capabilities, but this must be balanced with the strict data privacy and regulatory considerations that some customers face, which may favor on-prem deployments for the foreseeable future.
The platform’s development and go-to-market strategy reflect a careful balance between depth and accessibility. By providing a wide range of components and a path toward a production-oriented deployment, Cake aims to appeal to teams that have both the technical chops to operate complex AI infrastructure and the governance discipline required by larger organizations. This includes entities that are investigating RAG deployments, model-as-a-service integrations, and multi-component AI pipelines that combine data ingestion, labeling, retrieval, and model inference in a unified workflow.
Market context, competition, and the path toward hosted solutions
The evolution of Cake’s approach sits at the intersection of several established themes in enterprise software and AI infrastructure. First, the company’s emphasis on open source components aligns it with a broader industry trend toward leveraging community-driven tools while providing a robust enterprise-ready layer on top. The open source nature of the components—across data adapters, ingestion systems, labeling tools, databases, and AI APIs—presents both a strength and a challenge. On the one hand, enterprises value the transparency and innovation that open source affords; on the other hand, they seek a coherent, well-supported platform that manages complexity and provides reliability at scale. Cake’s model aims to offer exactly that: a curated, managed, and production-grade stack built from widely used open source components, with a focus on enterprise-grade security, governance, and operational controls.
In Europe, similar dynamics play out with players like Aiven, a data infrastructure company with a high profile and substantial capital backing. Aiven emphasizes cloud-native data services and enterprise-grade support for a broad set of open source data technologies. The comparison helps illuminate the market context in which Cake operates: a growing demand for enterprise-ready open source infrastructure that can be deployed with confidence, either on-premises or in the cloud, and that provides the necessary reliability and governance to support mission-critical AI workloads.
A parallel often drawn in the industry is the analogy to Red Hat, the archetypal enterprise-grade open source software distributor that IBM acquired for $34 billion and that became synonymous with providing secure, supported Linux for business use. The Red Hat model demonstrates how an enterprise can scale the adoption of open source technologies by combining community-driven innovation with professional-grade support, certification, and a robust ecosystem of partners. Cake’s leadership has invoked this comparison to articulate the company’s ambition: to replicate that same level of enterprise confidence for today’s AI stack, which now includes a broader and more complex set of open source components.
One important distinction in Cake’s positioning is its dual path of on-prem deployment with an eye toward eventual hosted offerings. The hosted version, when it arrives, could appeal to organizations with lower compliance obligations or those seeking operational simplicity and the benefits of a managed service. However, data privacy considerations and regulatory requirements make on-prem installations a necessity for many enterprises, particularly in regulated industries such as finance and healthcare. Cake’s current emphasis on self-hosted deployments reflects a pragmatic response to market demand, while the longer-term hosted option could unlock new segments and use cases.
From a competitive perspective, the open source AI infrastructure market remains fragmented, with a mix of startups and established players offering either narrow, project-level solutions or broader platforms. The value proposition of a curated, validated stack with integrated governance and security controls stands out in a landscape where integration complexity often undermines speed and reliability. Cake’s ability to demonstrate tangible performance gains, reduced time to deployment, and improved compliance posture will be critical as it scales its customer base and expands its ecosystem of components and partners.
In this broader market context, Cake’s traction signals—early customers, an experienced leadership team, and a notable seed investment from a high-profile venture fund—position the company to pursue a meaningful growth trajectory. The company’s strategic plan includes continuing to grow its user base, expanding its ecosystem of supported components, and pursuing additional financing in the near term to accelerate product development, go-to-market momentum, and the scale of its platform.
Traction, partnerships, and strategic direction for the next financing phase
Cake’s emergence into public view comes with an emphasis on its early customer footprint and the momentum it has achieved in fundraising. The company has publicly described working with customers in sectors where AI-driven data workflows can deliver meaningful business value, including AI bioscience and data intelligence within insurtech. The mix of customers, while modest in number, signals a practical focus on real-world deployments and the practicalities of integrating a broad set of open source components in production settings.
The seed round, led by Gradient Ventures with participation from Primary Venture Partners, Alumni Ventures, Friends & Family Capital, Correlation Ventures, and Firestreak Ventures, underscores the strategic interest from investors who see the potential of a platform that can simplify and secure open source AI infrastructure for enterprise use. The capital is intended to accelerate product development, broaden the platform’s component coverage, and strengthen go-to-market capabilities as Cake seeks broader adoption across industries that rely on robust, auditable AI workflows.
From a growth perspective, founder Herscu suggested that the company is already positioned as a Series A-type company in terms of traction, with the expectation that the Series A may carry characteristics more commonly associated with a Series B in terms of scale and market reach. This framing reflects an early realization of market validation and customer engagement that has begun to resemble a more mature stage of growth, even as the company remains in its seed-to-early-growth phase.
Looking ahead to 2025, Cake is positioning itself for continued fundraising to fuel continued platform development, expand its ecosystem, and accelerate customer acquisition. The founders’ background—ranging from their prior ventures to their deep engagement with AI executives across the data science and enterprise IT space—provides a narrative of seasoned execution and an understanding of enterprise adoption cycles. The seed funding and the strategic investor support from Gradient Ventures bolster this narrative, signaling to the market that Cake intends to move quickly to broaden its platform capabilities and extend its reach into more enterprise environments.
Leadership, technical strategy, and roadmap
The Cake leadership team blends deep engineering expertise with strategic acumen in venture-backed ventures. Misha Herscu, the CEO, brings a track record of building and exiting AI-focused ventures, alongside a hands-on approach to customer discovery and product-market fit. Skyler Thomas, the CTO, brings a background as a chief architect at IBM and as a distinguished engineer and director of strategy at Hewlett Packard Enterprise, where he was involved with MapR, a company later acquired by HPE. This combination of executive leadership and technical depth underpins Cake’s strategy to deliver a production-grade AI infrastructure built on open source foundations.
Thomas has repeatedly emphasized a practical caution about the enterprise adoption of AI technologies emerging from research labs. He notes that while a large number of open source tools are in circulation, the enterprise requires a more mature and secure environment—one that includes robust authentication and authorization mechanisms, as well as a dependable path to production. This perspective aligns with Cake’s focus on the "production-ready" dimension of its stack, ensuring that the curated components not only work well in isolated experiments but also remain reliable, auditable, and scalable in real-world enterprise deployments.
Thomas’s experience across IBM and HPE, including leadership roles in strategy and architecture, informs Cake’s perspective on how to design an integrated stack that reduces friction. He has observed a recurring pattern across projects: organizations can adopt open source tools, yet they often fail to achieve enterprise readiness because critical governance, security, and operational requirements are not addressed upfront. Cake’s approach is to address these shortcomings by delivering a cohesive platform that provides integration, governance, and security across a broad spectrum of open source components.
The roadmap, as articulated by the leadership, includes maintaining a self-hosted deployment model while keeping an eye on the feasibility and benefits of a hosted version. The decision to delay or delay-to-hosted options depends on customer demand, regulatory considerations, and the practicalities of data localization and privacy. The leadership emphasizes that for many customers, keeping data on their own premises remains non-negotiable due to privacy stipulations, but for others, a hosted solution could offer convenience, operational simplicity, and cost benefits. This nuanced stance allows Cake to serve a diverse set of customers with varying compliance and governance requirements.
In terms of product strategy, Cake’s emphasis on integrating and curating a broad set of open source components across the AI stack provides a unique value proposition. The platform aims to offer a unified operational model that streamlines deployment, simplifies maintenance, and reduces the time to value for complex AI workflows. The emphasis on security, authentication, and authorization, coupled with the scalability of the platform, positions Cake to address enterprise-grade demands for reliability and governance in AI deployments.
The company’s public narrative around the “big picture problem” also implies a longer-term product strategy that centers on the practical realities of enterprise AI adoption. By focusing on end-to-end integration, security, and production-readiness, Cake positions itself as a trusted foundation layer for organizations seeking to deploy advanced AI capabilities at scale. This strategic direction aligns with the market’s growing emphasis on responsible, compliant, and auditable AI infrastructure that can support mission-critical operations.
Conclusion
Cake’s emergence signals a focused effort to transform the open source AI infrastructure landscape into a production-ready, enterprise-grade platform. By aggregating and securing more than 100 components—across data sources, ingestion, labeling, databases, and generative AI interfaces—the company aims to reduce operational friction for teams seeking to deploy advanced AI workflows. The founding team’s background, the strategic backing from Google’s Gradient Ventures, and the early traction with clients in AI biosciences and data intelligence set a clear course for growth and expansion.
The company’s philosophy centers on solving the broad “big picture problem” of integrating diverse open source components into a cohesive, secure, and scalable stack. This approach seeks to deliver the benefits of open source—flexibility, innovation, and community-driven development—while providing the enterprise-grade reliability, governance, and support that organizations require for production deployments. As Cake advances toward its next financing round and consideration of a hosted offering, the market will be watching closely to see how the platform matures, how its ecosystem expands, and how effectively it translates this ambitious open source strategy into tangible value for enterprise AI initiatives.
