Dataiku

Dataiku support is exceptional. Their prompt response and thorough understanding of the underlying issue are un matched with other products. Dataiku offers a drag-and-drop development environment for faster development, but it also allows users to code in Python. This enables users to control every detailed aspect of components. In other words, Dataiku caters to both beginner users and advanced users who want to use scripts to automate tasks within the tool. Another important feature is the plugin ecosystem, which allows third-party plugins (mostly open-source) to extend the tools capabilities. Additionally, the LLM mesh and easy integration with most leading LLM libraries make the tool ideal for extending data science and ML projects with LLM capabilities. Dataiku offers a free community edition that allows teams to begin using the tool without any cost. This edition enables teams to assess the tools capabilities and determine if they require an enterprise-ready version. If so, they can choose from other available licenses.

A flow and set of actions that help users easily understand the data pipeline.

Dataiku's full suite of capabilities ranges from low-to-no-code visual recipes, multiple programming languages and coding environments (R, Python, SQL), visual application development choices (VS Code, Streamlit, Shiny, Dash, Webapp, Visual-as-Recipe etc.) to GenAI/LLMs/RAG, prompt engineering capabilities, etc. Such diverse tech offerings help meet the needs of various user personas, from technical to non-technical, from data scientists and advanced AI specialists to regular business users and AI consumers. All in a single environment. Dataiku also offers clients superior coaching, training and use case support network. All of this combined accelerates user adoption and greatly smooths the learning curve.

There are a few minor areas for improvement in the tools visual aspect. For instance, in the visual development environment (flow diagrams), users have limited control over manually arranging the visual recipe. Its all automatically arranged, which sometimes wed prefer to do differently. However, this lack of features doesnt affect the tools technical performance or capabilities. Another missing feature is the lack of integration with cloud platforms for deploying models as APIs. While the tool has Kubernetes infrastructure to serve API models, we prefer a GCP cloud-run environment. It offers better security, load-balancing capabilities, and a more cost-effective environment. The licensing is a bit expensive, particularly if an organization decides to onboard citizen data scientists from businesses using the tool. The developer license and potential return on investment may not be justifiable for some organizations.

The virtual training provided by the vendor to the company was difficult to understand and might discourage people from starting to use this product.

(1) Depending on each client's internal enterprise informatics and security infrastructure settings and domain specific compliance and constraints, the out-of-box Dataiku AI Platform often needs to be configured or set up to meet a client's specific needs. Therefore, it is important to ensure that the Dataiku AI Platform continues to provide a modularized and flexible architecture to meet client company's specific needs. (2) AI, especially the Agentic AI field is changing and evolving rapidly. There are many competing solutions out in the marketplace, we are still facing the challenge of which AI or Agentic AI platform to select that offers ease of integration into our enterprise ecosystem in terms of people, process and technology.