DataLab is a data notebook that (1) smartly leverages generative AI technology so you can ‘chat with your data’, (2) features a powerful IDE to review, tweak and run your analysis, and (3) seamlessly turns your work into an beautiful, shareable report.

Hey all! I'm Filip, leading the development of DataLab at DataCamp, and I'm super excited to show you what we've built in the past months and to hear your feedback. The next iteration of DataCamp Workspace, DataLab is an AI-powered data notebook to go from data to insights by chatting in a ChatGPT-like interface. Because Generative AI can still make mistakes, DataLab's AI assistant writes and runs code that you can seamlessly review, tweak and share, without leaving the tool. That way, you can fully trust the insights you're uncovering. DataLab can connect to any data, wherever it lives: CSV files, Google sheets, common data warehouses and databases are all supported. Finally, DataLab also has built-in collaboration and reporting capabilities, so you can go from data, to insight, to shared understanding in a matter of minutes. Give it a spin (for free!) and let us know what you think and how we can improve! https://www.datacamp.com/datalab Cheers, Filip

@max_savonin1 All great questions, let me answer them one by one > Can DataLab's AI assistant handle complex data transformations and preprocessing tasks within the chat interface? If you have a good idea of what transformations and preprocessing you want to do, I suggest you switch to DataLab's "Code View", which is a fully-featured notebook editor. It has an AI code generator built in to write code from scratch or extend some code you already wrote. > How does DataLab ensure the explainability and trustworthiness of the insights generated by the AI assistant? All the code that the AI Assistant in the chat generates, is being added to a corresponding notebook. Users can seamlessly switch to this code and review, tweak, run and share it. Hope this answers your question! Any shares in your network are greatly appreciated!

Oh wow, the video demo looks really great! ❤️ I think it's smart that you provide familiar UI to the users. It lowers friction. I can see that even non-technical people might try and generate reports themselves. Do you see non-technical users as a target audience for DataLab? All the best with the launch team DataLab 🚀

@jgani Thanks for your message! We're at this journey for quite a while already. We started out with a code-first data notebook, but were always looking for ways to make it more accessible and easier to use, so that anyone, regardless of technical skill, could get insights from their data. The rise of generative AI has provided a new way of going about this challenge, allowing users to describe in plain English what insight they want. The quality of large language models (LLMs) is not (yet!) at the level where you can just throw any data warehouse at it and expect it to give accurate answers in all cases, but it still unlocks a great analytics experience: > Before, without AI: Analyst spends hours building a dashboard that can answer every possible question colleagues might possibly have. Colleague with a question needs to dig through a bunch of filters to get the insight they need. > With DataLab: Analyst on the team prepares a clean dataset with all the required properties and data slices, and makes this available through DataLab. Colleague starts a DataLab chat, ask what they need, e.g. ‘how many subscriptions did we sell in the past 30 days in France’. DataLab generates and runs code in the background, and answers the question on the spot. The generated to answer the question can be easily inspected, adjusted and re-run in DataLab’s fully-featured data notebook. LLMs will get more powerful, and DataLab will become increasingly smart about how to get the best results (leveraging past interaction data, previous data queries, company documentation, etc), making life easier for the analyst and getting insights significantly less frustrating for less technical folks. This answer turned out longer than expected, but I'm pretty passionate about all this, hope you don't mind!

@filipsch I'm all too familiar with the 'before AI' world. We had to build a custom solution for non technical people, literally like you described it. DataLab is very well positioned to take advantage of the AI revolution. Congrats FIlip!

Just hit the upvote 'cause what you've got here sounds like a game changer. Chatting with data that's cool and kind of sci-fi. Plus, the collab IDE and snazzy report sharing sounds like you're making data talk turkey and look good doing it. Got a burning question though. How user-friendly is this for folks not super tech-savvy? Can a regular Joe get good stuff from this without a tech dictionary? Keep rocking it. Cheers!

@aman_wen Great question! In the "AI Chat" mode the user is not exposed to any code at all. Instead they just pop in a dataset and start asking questions. In the background, the AI is writing code to answer the question, but the user doesn't see this unless they want to. If the user decides to go to the "Code" mode, they're met with a clean, slick, yet powerful notebook editor with all the necessary features to delve deeper into the code, should they want to. Regardless of the mode that users are in, we're obsessed with keeping the UI and UX super simple and accessible, also for not super tech-savvy folks. Hope this answers your question!

The feature I most love is the seamless mixing of SQL and Python. I use Workspace to create reports of product metrics, querying from BigQuery and then visualizing results with Plotly!

Tried a Python course on DataCamp and was impressed by how smoothly everything worked. Explanations were sharp and easy to grasp, and hands-on tasks kept it engaging. Didn’t need extra tools, everything’s built-in. Worth checking out

This is awesome. Previously I was using Colab to do all of my note taking as well as sharing Jupyter Notebooks to discuss or ask for help. Now I have everything in one place!

DataLab

AI-powered data notebook

DataLab

DataLab Launches

Do you use DataLab?