As the year draws to a close, it’s time to roll out the red carpet for the standout product features of 2024. In classic awards-banquet style, we’ve given each feature a superlative that captures its personality, impact, and the value it brings to your data projects. From groundbreaking improvements to fan-favorite tools, these updates have made it easier for organizations to harness generative AI, reduce friction in their MLOps, and streamline day-to-day workflows. Without further ado, here are the winners of the Dataiku 2024 Feature Superlative Awards!
The Dataiku Awards: Celebrating 2024's Best Features
🏆 LLM Evaluation: The Perfectionist
For Always Demanding the Best
When it comes to LLM responses, good is good — but it could always be better. That’s why LLM evaluation is our Perfectionist. This feature holds every LLM response to the highest standard, meticulously evaluating model outputs with quantitative metrics like answer correctness, answer relevance, faithfulness, and context recall.
With Dataiku, evaluating and tracking LLM performance goes from a time-consuming, subjective process to a standardized, disciplined operation. The visual recipe brings clarity to design experiments and provides ongoing quality assurance and monitoring for production applications. With interactive row-level analysis, model comparisons, and the ability to automatically capture, store, and set up proactive alerts based on these metrics, this new feature is a real overachiever … in all the best ways!
Teams can iterate on LLM-powered applications with confidence, knowing they have the tools to track performance, compare metrics, and identify the best-performing configurations — all without writing a single line of custom code. For any organization scaling its use of GenAI, this is a game-changer.
🌎 Deploy Anywhere: The Diplomat
For Building Bridges to Everywhere
When it comes to ecosystem integration, the “Deploy Anywhere” capability in Dataiku is the ultimate Diplomat. This feature enables you to deploy models developed in Dataiku to third-party platforms like Databricks, Snowflake, AWS Sagemaker, Azure ML, and Google Vertex AI. There is no loss of control or visibility for Dataiku-native models once they are deployed to a cloud ML platform — all the same visualizations and explainability are present and ready to be shared with team members and stakeholders. Better yet, centralized model governance and unified monitoring dashboards provide comprehensive views of all deployed models and their statuses, so you can have centralized AI oversight across all your AI infrastructures.
By bridging the gap between Dataiku and external ML platforms, this feature elegantly connects disparate worlds in your tech ecosystem, democratizes model deployment to other platforms, and ensures smooth MLOps collaboration across teams … a true diplomat, indeed.
💡Dataiku Answers: The Know-It-All
For Always Having All the Answers
When you need information fast, simply turn to the Know-It-All. Dataiku Answers shines as the go-to for conversational AI use cases, whether you’re chatting with datasets, retrieving facts from approved knowledge banks using RAG techniques, or even generating brand new images on the fly. With a prebuilt front-end user interface and a point-and-click menu to configure the back end, this packaged feature makes it possible for teams to deliver a wide range of production-grade, chat-based solutions in weeks, not months.
Thanks to the Dataiku LLM Mesh, Dataiku Answers connects to your vector store and LLM providers of choice, so you have complete flexibility and future-proof control over which embedding, completion, and image generation models are used in your application. To foster trust in responses, the underlying LLM can source its answers from approved information in your knowledge bank and provide citations, extracted quotes, or the generated SQL query for complete transparency and explainability. End users can even upload their own documents for ad hoc analysis, making it the ultimate productivity hack. In short, it’s like having a brilliant colleague who’s always one step ahead.
✅ Data Quality: The Control Freak
For Keeping Everything in Line
It may not be the award we dream of winning as kids, but let’s face it: We all need a Control Freak on our team (and we love them for the rigor and discipline they bring to the table)! Data quality takes on that role with pride, streamlining data quality rules and monitoring health across projects.
From visual data quality indicators to rule templates to prebuilt dashboards, Dataiku’s suite of data quality features ensures your pipelines are always in tip-top shape. Whether you’re a data engineer or an IT operator, you’ll appreciate the automated vigilance and ability to catch small inconsistencies before they become big problems.
🗃️ Data Lineage: The Historian
For Keeping Track of the Details
While we’re on the topic, let’s stay on the data hygiene theme. Data must be updated as new columns, values, and analysis needs emerge, but in complex, multi-project initiatives where data is heavily interconnected, these updates can also lead to unintended downstream consequences. When it comes to understanding the root cause of changing data, history matters — and data lineage is here as our Historian to ensure it’s never forgotten. The data lineage feature in Dataiku helps you trace and understand the flow of data transformations across datasets and projects.
With column-level lineage views and a visual map, you can review exactly how a column’s data was modified and highlight where issues may have occurred. Impact analysis enables you to clearly see which systems, reports, or models rely on a particular column in a dataset. Whether you’re investigating data quality issues or assessing the impact of changes, data lineage provides the clarity you need to collaborate and troubleshoot effectively.
📋 Unified Monitoring: The Team Mom
For Looking Out for Everyone
Every team needs someone to keep things running smoothly, and that’s exactly why Unified Monitoring wins our Team Mom award. With its single-pane-of-glass interface in the Dataiku Deployer, this interactive dashboard provides consolidated oversight and status updates for batch jobs, API services deployments, and more. Acting as a central watchtower, it enables operators to monitor pipelines and models developed and deployed across diverse platforms such as Databricks, Snowflake, Azure ML, AWS Sagemaker, and Google Vertex AI.
The Team Mom keeps tabs on all your deployments in flight across diverse infrastructures, prioritizing warnings and errors so you can focus on what matters. Dependable, organized, and keenly aware of every item’s status — it’s the ops sidekick you never knew you needed.
🙌 Flow Controls & AI Assistants: The People-Pleaser
For Making Everyone's Life a Little Easier
When it comes to day-to-day data work, sometimes it’s the little things that bring the biggest smiles. With updates like manual Flow zone management, better experience when deleting or adding recipes in pipelines, and more AI assistants for in-app support, the People-Pleaser award is all about making life easier for users. Whether you’re tidying up your Flow or saving time on routine tasks, these features ensure that building Dataiku Flows is as friction-free and efficient as possible.
For instance, we currently have four embedded AI assistants (and more on the way next year) quietly waiting on standby to boost productivity and provide inspiration. Whether it’s generating data prep steps, debugging and documenting code, explaining Flows, or crafting the perfect recipe from a simple natural language description, these tools save time while elevating your final results. Like any good people-pleaser, these features anticipate your needs, smooth out difficulties, and provide support so you can focus on what matters most.
Wrapping It All Up
As the curtain closes on 2024, these highlighted features reflect Dataiku’s unwavering dedication to innovation and empowering users and we think this year’s innovations have something to delight everyone. Which feature stole the spotlight for you? While these are our standout features of the year, we know the real MVPs are the users who bring them to life. Thank you for a fantastic 2024, and here’s to reaching even greater heights in 2025!