A stealth-stage startup is stepping into the public eye with a mission to simplify how businesses assemble and operate open source AI infrastructure. Backed by Google’s AI-focused venture fund, the company, Cake, aims to consolidate and secure more than 100 enterprise-grade components—from data source adapters and ingestion tools to labeling services, vector and graph databases, and generative AI APIs—into a cohesive, production-ready stack. The goal is to reduce engineering overhead for organizations building and deploying AI applications, enabling faster time-to-value without compromising security or governance. By bundling diverse open source projects into a managed, ready-to-run platform, Cake seeks to transform how enterprises approach the modern AI stack, turning a sprawling ecosystem into a single, reliable solution for complex use cases.
The emergence of Cake: mission, scope, and core promise
Cake presents itself as a response to a persistent “big picture problem” faced by enterprises navigating the rapidly expanding AI tooling landscape. The essence of Cake’s approach is to harmonize a broad swath of open source components—across data ingestion, processing, storage, and AI tooling—into a curated, production-grade package. Rather than building a new monolithic AI product around a single project, Cake offers an integrated suite that spans the entire stack, designed to be deployed and maintained by in-house teams without sacrificing enterprise-grade reliability, security, or governance.
A practical way to understand Cake’s value is to look at how it handles the common, but complex, workflows that modern enterprises require. Consider a large financial services organization that possesses millions of documents containing intricate financial data. When such a company wants to perform retrieval-augmented generation (RAG) to enhance responses to natural-language queries, it often confronts two intertwined challenges: selecting compatible components and ensuring they work together securely at scale. If an off-the-shelf product cannot meet compliance requirements or lacks the flexibility to adapt to evolving needs, the firm may be forced to stitch together multiple disparate tools—an approach that is time-consuming, risky, and prone to maintenance burdens. Cake positions itself as the solution to this problem by providing a bundled, managed environment that brings the best open source components into a cohesive, enterprise-ready platform.
The concept of the “production-ready AI stack” at Cake is anchored in this philosophy: identify the critical layers that constitute modern AI infrastructure and offer a curated selection of proven, interoperable projects. This means not only offering components for data ingestion and transformation but also ensuring robust authentication, authorization, data privacy, and operational controls that enterprises expect. The ambition is to deliver a turnkey foundation that can be deployed with confidence, enabling teams to focus on delivering AI-driven outcomes rather than assembling and validating the underlying plumbing.
Cake’s product philosophy also emphasizes flexibility. While the company may eventually offer a hosted version, its current emphasis is on on-premises or controlled environments—consistent with many organizations’ data-privacy and regulatory requirements. The emphasis on controlled deployment aligns with the broader market reality: certain industries—healthcare, finance, government—often prefer or require keeping data behind their own firewalls or within tightly governed cloud boundaries. Cake’s approach thus prioritizes security, traceability, and governance as central features of its value proposition, not afterthoughts.
The company’s framing of the problem and its proposed remedy resonates with broader trends in the AI ecosystem. As AI tooling proliferates—from model zoos and data pipelines to specialized adapters and labeling tools—enterprises confront a proliferation of options. The promise of Cake is to translate this heterogeneity into a coherent, production-grade environment where teams can reliably build, test, and deploy AI solutions without rebuilding the wheel for every project. In this light, Cake’s value proposition extends beyond mere aggregation: it is about delivering a curated, stable, and auditable ecosystem that accelerates delivery while upholding enterprise-grade security and compliance standards.
Founders, funding, and early traction: building a venture around integration
Cake was founded in New York in 2022 by two technologists with deep experience at the intersection of AI, data, and enterprise software. Misha Herscu (CEO) and Skyler Thomas (CTO) have steered the company since its inception, guiding its product strategy and technical direction as Cake progressed from stealth to a formal public unveiling. The team reports that Cake has already begun working with notable customers, including an AI bioscience startup focused on drug discovery and a data intelligence insurtech, highlighting early validation across distinct verticals. While the company’s public profile has been relatively modest up to this point, its recent formal reveal marks a broader push to communicate its vision and market approach.
Financially, Cake has raised a total of $13 million since its inception. The breakdown includes $3 million in pre-seed funding—gathered during the company’s formative years—followed by a $10 million seed round led by Gradient Ventures, Google’s venture fund focused on AI initiatives. This seed round positions Cake to accelerate product development, deepen its partner ecosystem, and scale go-to-market efforts as it targets enterprise customers with complex AI infrastructure needs.
Herscu describes a disciplined, customer-centric approach to building the business. He notes that Cake has not been secretive so much as relentlessly focused on product development and customer work. The company’s leadership emphasizes that the work has centered on engagement with practitioners and decision-makers across organizations, rather than pursuing hype or speculative milestones. In reflecting on his journey, Herscu points to a prior track record in AI infrastructure and entrepreneurship as a foundation for Cake’s strategy, including his experience developing machine learning infrastructure for radiology in a previous venture and selling that company to an IT vendor in 2017.
Herscu’s background provides context for Cake’s emphasis on practical, enterprise-grade outcomes. Before founding Cake, he led McCoy Medical Technologies, a company that built AI infrastructure for radiology applications. The exit to TeraRecon—the IT vendor—gave him firsthand exposure to the challenges of deploying AI in regulated, mission-critical settings. After that exit, Herscu joined Primary Venture Partners in New York as an “operator in residence,” where he conducted extensive conversations with hundreds of data science and AI executives. He describes a deep, ongoing discovery process in which he identified a common set of pain points across enterprises: the difficulty of integrating a broad ecosystem of open source AI components into a single, reliable, production-ready platform.
The other founder, Skyler Thomas, has a rich background as well, including leadership roles in architecture and strategy for large technology organizations. Thomas’s experience spans roles at IBM as a chief architect and at Hewlett Packard Enterprise as a distinguished engineer and director of strategy. He also played a role in HP Enterprise’s acquisition of MapR, a significant milestone in the history of big data and AI infrastructure. Across his career, Thomas has observed a recurring pattern: enterprises increasingly rely on open source tools, yet the integration burden remains enormous. This insight continues to inform Cake’s approach, which seeks to deliver enterprise-grade integration with authentication, authorization, governance, and security baked into the platform from day one.
Both founders echo a shared narrative about the market’s current state: there is a thriving ecosystem of open source AI tools and data technologies, but the enterprise-grade delivery model that makes these tools usable at scale remains comparatively underdeveloped. They describe a world where enterprises must navigate a complex array of components—from data sources and ingestion pipelines to labeling services, vector databases, and generative AI APIs—and where the absence of a cohesive integration framework slows adoption and increases risk. Cake positions itself as the facilitator that makes such a framework affordable, reliable, and sustainable for teams that do not wish to become full-time integration engineers. This narrative is reinforced by conversations about the broader competitive landscape, including parallel efforts by other players in data infrastructure and open source ecosystems.
In terms of momentum, the seed round’s size and composition underscore investor confidence in Cake’s thesis: the idea that a curated, managed, open source AI infrastructure platform can unlock enterprise value by reducing integration toil and accelerating deployment timelines. The round also signals that the founders’ networks and prior experience are resonating with investors who understand both the technical and business dimensions of AI at scale. Looking ahead, Cake’s leadership acknowledges ongoing plans to pursue additional financing in the near to mid-term future, with an eye toward a larger round that could be interpreted as a Series A or its equivalent, depending on market dynamics and traction.
The big picture problem: why integration is the barrier and how Cake addresses it
The core challenge Cake references is not the lack of great open source AI components, but the arduous, often opaque process of stitching these components into a reliable, end-to-end stack that can be deployed and operated within enterprise constraints. In the eyes of Cake’s founders, most enterprises are confronted with a multi-layered reality: a lab environment full of experimental tools, a production environment with strict governance, and a deployment reality that demands consistent authentication, authorization, monitoring, and compliance.
This disconnect—between the potential of individual open source projects and the practicality of deploying them at scale—creates a spectrum of inefficiencies. Teams frequently spend substantial time evaluating, testing, integrating, and validating tools for compatibility and security. The broader ecosystem may evolve rapidly, with new releases, patches, and features requiring ongoing maintenance and revalidation. The time and cost associated with keeping a production AI infrastructure up-to-date across dozens of components can be significant, diverting engineering resources from delivering business value.
Cake aims to address these pain points by offering a curated, production-ready bundle of open source AI infrastructure. The company’s approach blends several essential elements:
-
Curation and interoperability: Rather than relying on a mix-and-match approach, Cake select s proven components across the AI stack and ensures they can be deployed together with consistent security and governance baked in. The goal is to minimize compatibility frictions that typically arise when teams combine disparate tools sourced from different development communities.
-
Managed delivery: Cake positions itself as a managed provider for open source AI infrastructure. This means customers receive a ready-to-run platform that handles updates, maintenance, and operational support, reducing the need for in-house teams to become experts in every tool. This managed angle is designed to accelerate adoption by reducing the risk and complexity associated with DIY integration.
-
Enterprise-grade security and governance: Given the sensitivity of enterprise data and the regulatory environments in which many AI applications operate, Cake emphasizes authentication, authorization, access control, data privacy, auditability, and compliance. These features are not add-ons; they are core attributes of Cake’s platform, embedded in the architecture from the outset.
-
Production-readiness and reliability: The platform is designed to support production workloads, including stringent performance, reliability, and observability requirements. Enterprises need robust monitoring, tracing, fault tolerance, and disaster recovery capabilities, and Cake frames these capabilities as integral to its offering rather than optional extras.
-
Data privacy and on-prem deployment orientation: Recognizing that many customers prefer not to move sensitive data to external environments, Cake’s current model emphasizes on-prem or tightly controlled deployments. While a hosted version is on the horizon, the initial focus is on keeping data within customer premises or governed cloud environments to align with privacy and compliance needs.
The practical implications of this approach are extensive. For a large financial services organization, the ability to deploy a vetted, production-grade stack that includes data adapters, ingestion components, labeling tools, and vector or graph databases reduces the time to implement RAG-based solutions and other AI features. It also reduces risk by ensuring that the integrated system adheres to enterprise security standards, reducing the likelihood of misconfigurations that could expose data or introduce vulnerabilities. In healthcare contexts, such as secure analysis of CT scans, a similarly integrated approach can streamline workflows while preserving patient privacy and regulatory compliance.
Cake’s emphasis on a “big picture” integration challenge reflects a broader shift in the AI space: as capabilities expand, so does the need for stability, governance, and operational maturity. Enterprises increasingly demand platforms that can absorb rapid innovation without introducing chaos into production environments. By providing a curated, managed, end-to-end stack, Cake seeks to enable organizations to leverage the richness of open source AI tooling while avoiding the fragmentation that historically slowed enterprise adoption.
Parallel development and enterprise open source adoption: the technical and strategic backdrop
Cake’s technical leadership brings a perspective shaped by years of work at major technology organizations. The CTO, Skyler Thomas, previously held senior engineering and strategy roles at IBM and Hewlett Packard Enterprise. He led initiatives in which the company engaged with complex data workloads and large-scale software infrastructure. Thomas’s career includes involvement with MapR through an HP Enterprise pursuit, highlighting exposure to the kinds of big data and analytics platforms that underpin modern AI workflows. Throughout his career, Thomas has observed a recurring theme: even when cutting-edge open source tools are compelling in a lab setting or for niche pilots, enterprises require more than raw innovation; they require a mature ecosystem around those tools—one that offers security, governance, operations, and a reliable support model.
Thomas’s reflections underscore a broader market dynamic: while the open source AI and data infrastructure space is expansive and vibrant, the path to enterprise deployment remains arduous. Many tools arrive with impressive capabilities but lack essential enterprise attributes such as scalable authentication frameworks, role-based access control, multi-tenant security guarantees, and production-oriented documentation and support. The enterprise community often needs not only the software itself but a dependable integration and service model that can be trusted in high-stakes contexts. Cake’s value proposition, in this sense, is to bridge the gap between the rapid evolution of research-grade tools and the stringent requirements of enterprise-grade deployment.
In drawing parallels to existing players in the data infrastructure space, Cake’s leadership points to notable European and global efforts that pursue similar objectives in different ways. For example, Aiven, a Finnish data infrastructure company valued at several billions of dollars, represents a parallel approach in which the focus is on delivering reliable, enterprise-grade data services. The comparison to Red Hat is particularly instructive: Red Hat’s rise to prominence and eventual acquisition by IBM for tens of billions of dollars is often cited as a blueprint for how to bring open source software into a large enterprise through a robust support and governance model. In Cake’s storytelling, these analogies are invoked to illustrate the potential path for AI infrastructure—one in which open source innovation is made enterprise-safe through a combination of certification, long-term support, and an integrated platform approach.
A relevant historical analogy helps frame Cake’s ambition: in the early days of Linux, thousands of open source packages competed for attention but lacked a secure, user-friendly support model. The emergence of Red Hat filled that vacuum by offering an enterprise-grade distribution with a reliable update mechanism, commercial support, and a governance model that enterprises could trust. Cake is positioning itself as the AI-era counterpart to that story—an organization that curates and stabilizes a broad AI toolchain to make it consumable by enterprises without forcing them to become open source experts themselves. This framing is more than rhetorical; it reflects a strategic focus on delivering reliability, governance, and an ecosystem of services that make a heterogeneous AI toolkit ready for production use.
On the deployment model, Cake acknowledges that a hosted version would be convenient for many organizations, but stresses that the current choice to emphasize on-prem and controlled environments aligns with data privacy and regulatory realities faced by customers. The company suggests that for some customers, cloud-hosted approaches bring additional complexity and risk, whereas maintaining control over the data lifecycle within trusted environments can simplify compliance and governance. Nevertheless, Cake’s leadership indicates openness to a hosted offering in the future, recognizing that a hosted model could broaden accessibility for teams with less stringent compliance requirements or with different cloud strategies.
The funding structure behind Cake—led by Gradient Ventures, with participation from Primary Venture Partners and others—reflects a broader investor interest in platforms that can scale AI at the enterprise layer. Gradient Ventures’ backing signals confidence in the team’s ability to translate open source AI innovation into enterprise-grade value, while the collaboration with an established early-stage partner network supports Cake’s go-to-market and architectural efforts. The funding combination also suggests a strategic intent to leverageGradient’s network for partnerships, customer introductions, and assistance with governance and risk management in regulated industries.
Market context: positioning Cake within a dynamic enterprise AI landscape
Cake operates in a rapidly evolving market where AI tooling spans data infrastructure, model deployment, data governance, and developer tooling. Several macro trends shape the company’s market positioning:
-
Open source as a foundation for AI: Enterprises increasingly rely on open source components to avoid vendor lock-in and to tailor capabilities to specific use cases. However, the deployment and operationalization of open source AI stacks require substantial engineering discipline. Cake’s value proposition centers on turning a scattered ecosystem into a cohesive, enterprise-ready platform.
-
The complexity of production AI: AI pipelines in production are not simply about model quality. They involve data pipelines, monitoring, security, privacy, reproducibility, audit trails, and governance. Cake’s platform emphasizes these operational aspects, aiming to reduce the risk and overhead associated with running AI in production environments.
-
The tension between on-prem and cloud hosting: Data privacy and regulatory compliance strongly influence deployment decisions. While cloud-based solutions offer scalability and ease of management, many enterprises prefer or require on-premises or tightly controlled environments. Cake’s approach acknowledges this reality and builds its product accordingly, while leaving open the possibility of a hosted offering in the future.
-
Competitive landscape: The market includes data infrastructure players that offer hosted services, distributed system tooling, and enterprise-grade support for open source stacks. Comparisons to Red Hat-like models for AI infrastructure illustrate a broader narrative: open source ethics and community-driven innovation combined with enterprise-grade governance and services can yield scalable market leadership.
-
The maturation of AI workflows: Enterprises are moving beyond pilots to real-world deployments that touch data governance, security, and regulatory concerns. This shift creates demand for platforms that reduce integration friction, accelerate delivery, and provide robust operational capabilities—precisely the area Cake aims to address.
In this landscape, Cake’s emphasis on curated, production-ready integration of open source AI components—paired with a governance-first approach and a clear focus on enterprise deployment realities—positions it as a compelling option for organizations seeking to modernize their AI infrastructure without sacrificing control or compliance. The seed funding and high-profile backers underscore the market’s appetite for solutions that can operationalize AI at scale, while the founders’ emphasis on customer-centric development suggests a strategy grounded in real-world needs and use cases.
Roadmap, hosted offerings, and the path to scale
Cake has signaled plans that include both continued on-prem deployment and the potential introduction of a hosted version in the future. The current stance is anchored in a belief that many organizations require strict data-location controls and privacy protections, making on-prem or tightly governed environments the most straightforward path to widespread adoption. However, the team recognizes that a hosted model could unlock broader access for teams with lighter compliance regimes or different organizational constraints. The decision to delay or phase in a hosted offering reflects a careful balance between customer needs, data governance considerations, and the operational challenges of managing a multi-component AI stack at scale.
From a funding perspective, Cake’s leadership has indicated readiness for further rounds as traction matures. There is an expectation that the company will pursue additional financing around the middle of 2025, aligning with a broader assessment of product-market fit and growth potential. The founders describe the early traction as indicative of Series A-grade momentum, suggesting that subsequent rounds may resemble a Series A or even potential Series B depending on how quickly the company scales its customer base, expands its partner ecosystem, and enhances its platform capabilities. This forward-looking stance signals a plan to invest in product expansion, ecosystem development, and international expansion where appropriate, all while maintaining a core focus on reliability, governance, and enterprise-grade performance.
The strategic emphasis on enterprise adoption is complemented by a focus on expanding use cases across industries. Cake’s messaging highlights scenarios ranging from financial services’ need for robust RAG-enabled document processing to healthcare’s demand for secure image analysis and e-commerce’s desire for stronger recommendation engines. By describing a spectrum of real-world applications, Cake anchors its platform in tangible business value rather than abstract capabilities. The company’s approach to roadmapping emphasizes modularity and extensibility, ensuring that the platform can accommodate new components and capabilities as the AI ecosystem evolves. In practical terms, this means a stable core platform with a flexible extension model, enabling customers to adopt new tools without destabilizing their existing production environments.
Use cases and industry applications: translating capability into outcomes
Cake’s platform is positioned to support a wide range of industry use cases that exemplify its core value propositions:
-
Financial services and RAG workflows: Enterprises with vast document collections can leverage the integrated AI stack to perform retrieval-augmented generation, enabling more precise and context-aware responses to natural-language queries. By leveraging curated data adapters, ingestion pipelines, and secure model interfaces, these organizations can extract insights, summarize documents, and respond to inquiries with auditable provenance and governance.
-
Healthcare imaging and diagnostics: Hospitals and clinical institutions require secure processing of medical images and patient data. An integrated AI infrastructure can enable advanced analytics on imaging data, support radiology workflows, and maintain strict privacy controls. Cake’s model emphasizes secure data handling, access control, and reproducible workflows, which are critical in regulated healthcare environments.
-
E-commerce and customer experience: Online retailers seek to improve recommendations, search, and personalization. An enterprise-grade AI stack can underpin advanced recommendation engines, optimized search relevance, and personalized user journeys. By integrating data sources, labeling pipelines for feedback loops, and scalable embeddings-based retrieval, Cake can help retailers deliver more accurate and timely insights to customers while maintaining governance and security.
-
Data science enablement and automated ML pipelines: Beyond vertical-specific use cases, Cake positions itself as a facilitator for data science teams to rapidly assemble, test, and deploy AI workflows. The platform’s emphasis on production-readiness and governance makes it a compelling foundation for teams seeking to operationalize experimentation and scale successful models.
-
Regulatory compliance and risk management: Enterprises face ongoing obligations to demonstrate auditable data handling and model governance. Cake’s platform can provide the controls and traceability required to meet regulatory expectations, including access management, data lineage, and change management across the AI stack.
The breadth of potential applications underscores how Cake’s approach aims to reduce the time and effort involved in moving from concept to production. By providing a production-ready, integrated platform that covers the major layers of the AI stack, Cake offers a practical path for organizations to realize AI-driven outcomes across diverse contexts while maintaining the governance and security requirements that matter most in enterprise environments.
Leadership, culture, and the path forward
Cake’s leadership emphasizes a pragmatic, customer-first ethos that has guided the company from stealth to public launch. The founders describe a collaborative, discovery-driven culture in which early customer engagement informs product direction and priorities. This approach is consistent with their prior experiences in startup and enterprise settings, where understanding the real-world constraints of customers is essential to building scalable platforms.
The leadership team’s background—combining deep technical experience with a focus on enterprise delivery—shapes Cake’s trajectory. The CTO’s history of working on large-scale data and AI infrastructure, combined with the CEO’s entrepreneurial background and commitment to customer discovery, positions Cake to navigate the complexities of enterprise adoption. As the market for AI infrastructure evolves, the company’s strategy is to maintain a balance between open source excellence and enterprise-grade governance, ensuring that the platform remains robust, secure, and adaptable.
In terms of go-to-market strategy, Cake appears to be leaning on a combination of direct engagements with enterprise customers and partnerships with systems integrators and technology partners who understand the nuances of regulated industries. The collaboration with Gradient Ventures, a fund known for supportingAI-centric startups, provides not only capital but also strategic guidance and an ecosystem of relationships that can help Cake expand its footprint in key markets. This combination of product focus and strategic network could prove instrumental as Cake scales its team, builds out its platform capabilities, and expands its customer base across industries.
The broader message from Cake’s leadership is one of confidence in the platform’s ability to unlock real business value by simplifying the complexity of the AI stack. The company’s narrative emphasizes that the challenge is not merely assembling tools but delivering a trusted, production-ready foundation that enterprise teams can rely on. In this framing, Cake is less about chasing the latest single technology and more about delivering a sustainable platform that can absorb continuous AI innovation while maintaining operational discipline.
Conclusion
Cake has unveiled a bold initiative to transform how enterprises build and operate open source AI infrastructure. By integrating a broad spectrum of components into a curated, production-ready platform, the company seeks to reduce engineering overhead, accelerate deployment, and provide governance and security at scale. With backing from Gradient Ventures and a track record of founder experience in AI and data infrastructure, Cake positions itself as a practical, enterprise-focused solution in a landscape characterized by rapid tooling evolution and rising demand for reliable production capabilities.
The founders articulate a clear vision: move beyond piecemeal assembly toward a cohesive stack that can be deployed with confidence in regulated environments. This includes a commitment to on-prem or tightly controlled deployments, with a hosted option anticipated in the future as customer needs and market conditions evolve. The company’s narrative emphasizes the “big picture problem” of integration across hundreds of open source components and the necessity of a managed, curated approach to unlock real business value from AI investments.
As Cake continues its journey toward scale, its trajectory will likely hinge on its ability to deliver on its promise of reliability, security, and practical outcomes for enterprise customers. If the company can demonstrate measurable improvements in time-to-production, cost of ownership, and governance compliance across diverse use cases—from finance and healthcare to e-commerce and data science—Cake could emerge as a foundational platform in the enterprise AI landscape, enabling teams to innovate with speed while maintaining the controls that matter most to business and regulatory stakeholders.
