ZenML aspires to act as the unifying force for various open-source AI tools, enabling data scientists, machine-learning engineers, and platform engineers to collaborate and construct new and Custom AI models through pipelines. What makes ZenML intriguing is its potential to empower companies to create their own customized models, reducing their reliance on API providers like OpenAI and Anthropic.
Louis Coppey, a partner at VC firm Point Nine, envisions ZenML as a tool that will come into play once the initial excitement of using OpenAI and closed-source APIs subsides, allowing people to construct their unique AI stacks.
Custom AI Models : Develop ML pipelines for companies
Earlier this year, ZenML secured an extension of its seed round funding from Point Nine, along with participation from existing investor Crane, bringing its total funding to $6.4 million since its inception.
The founders of ZenML, Adam Probst and Hamza Tahir, initially worked together on a venture that developed ML pipelines for companies in a specific industry. Their experience led them to design a modular system adaptable to diverse circumstances, environments, and customer needs, culminating in the creation of ZenML. This system, referred to as MLOps, streamlines machine learning operations, akin to how DevOps functions in software development.
ZenML’s core concept revolves around pipelines. These pipelines can be written and executed locally or deployed using open-source tools like Airflow or Kubeflow. The framework integrates with various open-source ML tools, including those from Hugging Face, MLflow, TensorFlow, and PyTorch, offering a unified, multi-vendor, multi-cloud experience with connectors, observability, and auditability for ML workflows.
Initially released as an open-source tool on GitHub, ZenML has gained substantial popularity, with over 3,000 stars on the platform. The company has also launched a cloud version with managed servers, with plans to introduce triggers for continuous integration and deployment (CI/CD).
ZenML has found applications in diverse sectors, including industrial use cases, e-commerce recommendation systems, and medical image recognition, with clients such as Rivian, Playtika, and Leroy Merlin.
ZenML’s success depends on the evolving AI ecosystem. While many companies currently rely on APIs like OpenAI’s for AI features, these APIs are often too sophisticated and costly for specific use cases. ZenML believes that the future of AI will involve companies developing their own solutions, and open-source and Custom AI Models are an attractive option for this purpose.
Ethical and legal considerations surrounding AI usage are also evolving, with European legislation encouraging the use of AI models trained on specific datasets and for specific purposes.
Gartner’s prediction that 75% of enterprises will shift from proofs of concept to production in 2024 underscores the growing importance of AI in business. ZenML envisions a future where specialized, cost-effective Custom AI Models, and smaller in-house models drive 99% of AI use cases.
