Vanna is an AI-powered data exploration and analytics platform that helps users to understand their data and make better decisions. It offers a variety of features, including:

  • Data exploration: Vanna allows users to explore their data by filtering and sorting it, and by creating custom visualizations.
  • Data analysis: Vanna provides users with a variety of tools for analyzing their data, including machine learning algorithms and statistical models.
  • Data sharing and collaboration: Vanna allows users to share their data and collaborate with others on data analysis projects.
  • Data security and privacy: Vanna encrypts all user data at rest and in transit, and it provides users with granular controls over who can access their data.

 

Overall, Vanna is a powerful and versatile data exploration and analytics platform that can be used by a variety of users to gain new insights from their data.

Chris: Hi there, please introduce yourself.

Zain: Hello, my name is Zain Hoda, Co-Founder of Vanna AI. I’m also the former founder of Alpha Hat, an alternative data analytics company focused on mobile geolocation data, which was acquired by Earnest Analytics. Before venturing into entrepreneurship, I worked in quantitative analysis for various financial firms, including BlackRock.

Chris: What inspired the creation of Vanna, and what specific challenges or needs in the field of data exploration and analytics did you aim to address with this platform?

Zain: The inspiration behind Vanna was an experiment that blossomed into both an open-source Python project and a product. Throughout my career in data analysis, I’ve been fascinated by the potential of technologies like ChatGPT. I began experimenting to see if it could generate SQL queries, and to my delight, it could—quite effectively, especially with the right prompts and reference queries. This capability is particularly useful for analysts like myself who, when faced with ad-hoc questions, would typically modify existing complex SQL queries. Language models are ideally suited for this task, as they can infer a great deal from such reference queries.

Chris: Data exploration is a fundamental step in data analysis. Can you elaborate on the tools and features that Vanna provides to allow users to filter, sort, and create custom visualizations from their data?

Zain: Vanna enables users to ask any question of their data, facilitating deep dives during data exploration. For example, an e-commerce company might ask a range of questions about searches, conversion rates, bounce rates, and cart abandonment. Each question generates an SQL query, tabular outputs, and an AI-created visualization, which can be shared with colleagues. This collaborative aspect is enhanced within platforms like Slack, where a thread of inquiries can lead to rich insights.

Chris: Data analysis is a broad area that often involves complex algorithms and models. How does Vanna make machine learning algorithms and statistical models accessible to users, and can you provide examples of scenarios where these have proven valuable?

Zain: Vanna is often seen as a tool that could replace employees, but it’s primarily used by data scientists looking to offload routine query answering. This frees them to focus on developing machine learning and statistical models. Once these models are operational and accessible through SQL stored procedures, such as with Snowflake’s Snowpark, they can be leveraged by anyone in the organization simply by posing questions to the AI.

Chris: Data security and privacy are paramount, especially when handling sensitive data. Could you provide more details about the encryption and access controls that Vanna employs to protect user data?

Zain: Data security and privacy is the number 1 reason that our codebase is open-source. You can actually choose to run every single component of Vanna locally and we do see certain industries like finance and healthcare where data teams are choosing to deploy everything themselves.
The paramount importance of data security and privacy is precisely why our codebase is open-source. Users can opt to run Vanna entirely on-premises, which is a common choice in industries like finance and healthcare. Initial use often occurs through a Jupyter notebook, with database connections remaining local so that Vanna does not access database details or contents. For broader deployment, users can run their web app, Streamlit app, or Slackbot independently or opt for our hosted solutions.

Chris: Business users can benefit from data analysis as well. What specific features or functionalities does Vanna offer to help business users make better decisions and improve marketing campaigns, for example?

Zain: While our initial users tend to be data analysts, business users are usually the final target audience because they’re the ones with the most questions. By using AI to get your questions answered, it allows you to shorten the feedback loop of getting follow-up questions answered. In fact, once you ask Vanna a question, it will automatically generate suggested follow-up questions that you can just click on to dive deeper into the data. In marketing in particular, what we’re seeing is that Google Analytics 4 has been very confusing for a lot of folks. We’re seeing a lot of data transfer from Google Analytics to Snowflake or BigQuery to make the data easier to analyze. If you’re doing that using Airbyte or Fivetran, we have pre-trained models that already understand the structure of Google Analytics and Google Ads data so that you can begin asking questions immediately.

Chris: In a constantly evolving field like data analysis, how does Vanna plan to remain innovative and continue meeting the changing needs and expectations of users?

Zain: Vanna thrives on user contributions as an open-source project. Users with urgent feature requests can add them directly, and we are currently expanding our developer relations team to further support and accelerate this process.

Chris: What is the pricing model for Vanna, and how does it compare to other data exploration and analytics solutions in the market? What value does it offer to users for its cost?

Zain: Vanna provides a free tier where we manage the infrastructure and LLM calls. The software itself, being open-source, carries no cost if you supply your own LLM. Customized deployments for enterprises, which may require integration with existing authentication systems, are where we typically charge.

Chris: Can you share success stories or user testimonials that demonstrate the significant impact Vanna has had on users’ data analysis and decision-making processes?

Zain: Here are some quotes from our users:

“It’s incredible how good it works, doing so with langchain is terribly awful ! also tried to do so with preprompting to give more context ect… when i used to work with data in memory and it never worked this good !”

“My hackathon project was a big hit today, so much so that I already have had conversations with several other teams that are interested. You have a really awesome product and I am looking forward to really exercising it. ”

“I want to thank you for the information provided; it has been very helpful. The test was successful and my superiors now want to see the application integrated with Slack”

Chris: Looking ahead, are there any exciting updates, new features, or developments in the pipeline for Vanna that users can look forward to?

Zain: We’re excited about a fully hosted solution that’s nearing release. It will allow users to seamlessly transition between Jupyter notebooks, web apps, and bots for Slack or Microsoft Teams, all without requiring any self-deployment.

Chris: Thanks for being with me, any last words? Where can our readers follow you?

Zain: Thank you for having me! I welcome readers to connect with me on LinkedIn.