Today we’re excited to officially launch Einblick Prompt, our context-aware AI agent. This project has been months in the making. With Prompt in every canvas, our users can now generate entire data workflows with as little as one sentence. More specifically, users can:
- Generate code with just natural language prompts–and run, debug, or edit immediately in-product
- Fix code with a click, without tab-switching to Google or Stack Overflow
- Create beautiful and complex graphs quickly
- And so much more.
Since its inception in the research labs of MIT and Brown University, the team has been dedicated to creating the most optimal workspace for data scientists and data teams. To that end, over the past year, the team has been working to bring the flexibility of generative AI’s natural language capabilities directly into our data notebooks. We strongly believe that the future of code is natural language–not as a replacement for programming, but as a tool that increases the efficiency of the programmer.
Solve any data problem with just one sentence
Our goal when building Einblick Prompt was to see if generative AI and natural language could truly speed up the creation of data workflows for data teams. Powered by OpenAI and LangChain, Prompt can help our users kickstart projects, generate code, fix code, and much more. Here are some prompts you can try out:
Speed up tedious data cleaning tasks
- “Join table1 and table2 on col1 = col2.”
- “Filter data for condition1 in col1 and condition2 in col2.”
- “Create a new column for new_var based on var1 and var2.”
Create complex graphs to present to stakeholders
- “Create a 2 x 3 grid of histograms showing the distribution of col1 by col2 and col3.”
- “Create a scatterplot of var1 and var2, color the points by var3.”
- “Recreate the above plot in seaborn using a pastel color palette."
Build ML models and NLP tasks
- “Use the vader package to do sentiment analysis of col1.”
- “Predict var1 using logistic regression.”
- “Build a linear regression model predicting var1.”
Why we built Einblick Prompt
When we were thinking about how to harness the power of generative AI and LLMs for data tasks specifically, we had a few requirements for our ultimate product:
- Address the pain points people had experienced using existing generative AI apps. (i.e. copy-pasting code, instability, inaccuracies, etc.)
- Create a context-aware tool to lessen the burden of prompt engineering on users.
- Go beyond code completion and text-to-code.
- Create a tool that our team and other data teams would love.
As many of us have experienced using tools like ChatGPT, there is a level of unpredictability when utilizing generative AI in apps. (In part this is because when ChatGPT was released in winter 2022, it had a knowledge cutoff of 2021). Prompt engineering–how to write the natural language requests just so–is a huge topic right now.
We wanted a tool that our team would love using to accomplish their day-to-day tasks. Tools like LangChain and AutoGPT had popularized ways to make generative AI and large language models (LLMs) smarter and more responsive, by focusing on AI agents that were context-aware and could reason alongside the user. This technology excited us, and we wanted to apply the concepts of reasoning agents to the data analytics and data science domain.
Get started with Einblick Prompt
To start using Prompt, all you need to do is:
- Sign in to your Einblick account
- Open a canvas
- Upload or connect to your data
- Right-click anywhere in the canvas > Prompt
- Type in your prompt
- Try "Create a scatterplot of var1 and var2."
FAQ
You can sign up for a free Einblick account at any time. As of July 2023, Prompt is available to all Einblick users.
At this time, we have released the feature to all Einblick users, including those on the free tier, at no extra charge. Rate limits for free users may be introduced.
Yes! You can sign up for a free Einblick account any time. Check out our quick start guide and demo video for more information.
Your datasets are not being shared with any third parties, but workflow metadata, including column names, may be shared to provide context. Furthermore, we only partner with vetted third parties who comply with industry best practice security policies, and have audited proof of adherence to standards like SOC 2.
We have extended OpenAI’s API with additional training done over data, Einblick Cells, and workflow-specific context. That’s why you can ask Einblick Prompt to build a workflow over your data, but ChatGPT by default would have no idea.
You do not. Einblick and all its features, including Prompt, are cloud hosted and accessed via your browser.
About
Einblick is an AI-native data science platform that provides data teams with an agile workflow to swiftly explore data, build predictive models, and deploy data apps. Founded in 2020, Einblick was developed based on six years of research at MIT and Brown University. Einblick is funded by Amplify Partners, Flybridge, Samsung Next, Dell Technologies Capital, and Intel Capital. For more information, please visit www.einblick.ai and follow us on LinkedIn and Twitter.