How does Quadratic handle large datasets and performance?

Quadratic is built for high-performance, supporting millions of rows. It uses WebAssembly for running Python and JavaScript within the browser and WebGL for rendering the sheet, ensuring fast response times even with large datasets.

Can I use traditional formulas in Quadratic?

Yes, Quadratic supports traditional spreadsheet formulas. You can also switch to coding in Python or executing database queries directly within the sheet, offering a flexible environment for data manipulation and analysis.

How does the AI in Quadratic get context about the spreadsheet?

Quadratic provides the AI with a summary of the entire sheet and the ability to access details as needed. It understands the location of tables and code cells and can use tool calls to perform actions, aiming to enable the AI to see and do anything a user can do.

What are the benefits of using Quadratic for data analytics?

Quadratic is ideal for analytics, helping users transform, visualize, and understand data. It enables collaboration between technical and non-technical team members on dashboards and analysis, filling a gap often seen in other BI tools.

Can Quadratic be used for proactive analysis or by other AI agents?

Yes, Quadratic is designed to be a tool that other agents might build upon. Its spreadsheet interface is considered the best for humans to understand and interact with data, making it suitable for agents to generate spreadsheets to answer user questions or for proactive analysis when data anomalies are detected.

What is the future of formulas versus Python in Quadratic?

The speaker anticipates formulas will be used less in favor of Python for complex data aggregation and analysis. While simple formulas for summing columns are still useful, Python offers better readability and power for intricate queries, especially when generated by AI.

How does Quadratic handle data connectivity?

Quadratic offers robust data connectors, supporting MySQL, Microsoft SQL Server, Snowflake, and PostgreSQL, with plans to add BigQuery, Neon, and Superbase support soon. This allows users to connect directly to their databases for analysis.

Key Moments

Quadratic: The AI Spreadsheet

Latent Space Podcast

Science & Technology4 min read29 min video

Jun 10, 2025|1,866 views|41|4

Save to Pod

Key Moments

TL;DR

Quadratic integrates AI and native Python into spreadsheets for intuitive data analysis and visualization.

Key Insights

Quadratic offers a next-generation spreadsheet experience that natively supports Python and database connections.

AI integration significantly lowers the learning curve for users, enabling them to perform complex data analysis and visualizations through natural language prompts.

The platform supports millions of rows and offers familiar spreadsheet interactions, enhanced with pinch-and-zoom capabilities and a unique Figma-like UX.

Users can combine traditional formulas, Python code, and database queries within the same sheet, with all execution happening client-side via WebAssembly.

Quadratic prioritizes real-time collaboration and provides a unified interface for both technical and non-technical team members in data analytics.

The AI assistant provides context-aware analysis, can access and manipulate data within the sheet, and learns from specific spreadsheet data interactions.

FROM TECHNICAL SPREADSHEET TO AI-POWERED TOOL

Quadratic initially focused on building a high-performance, next-generation spreadsheet that natively speaks Python and connects to databases. This technical foundation served a group of early, more technical adopters. The recent integration of AI has dramatically broadened its appeal, bridging the gap for users who are not proficient in coding or complex formulas. This AI layer acts as an intuitive assistant, allowing anyone to analyze and visualize data simply by using natural language prompts, effectively zeroing out the learning curve and boosting adoption.

INTUITIVE USER EXPERIENCE AND INTERACTIVE ANALYSIS

The user interface of Quadratic is designed to feel immediately familiar, mimicking the look and feel of traditional spreadsheets. The platform supports millions of rows, allowing for large dataset handling. A key UX enhancement is the pinch-and-zoom functionality, providing a dynamic way to navigate and focus on any part of the data, akin to a Figma-like experience. Users can seamlessly drag and drop CSV files and then interact with an AI chat panel on the side to orchestrate analysis, generate visualizations like revenue-over-time charts, and even apply transformations such as logarithmic scales directly through prompts.

INTEGRATED FORMULAS, PYTHON, AND DATABASE CONNECTIVITY

Quadratic empowers users by allowing them to blend traditional spreadsheet formulas (accessed via '=') with native Python code execution (via '/') and direct database queries within the same interface. Python code runs in a WebAssembly container within the browser, ensuring fast response times. For instance, a Python cell can output data directly into the sheet as a table, or a database query can pull information based on natural language prompts, recognizing the schema context. This unification of data sources and computation methods within a single, familiar spreadsheet interface offers unparalleled flexibility for data manipulation and analysis.

CLIENT-SIDE EXECUTION AND COLLABORATIVE DATA MODELING

All computations, including Python code and database queries, are executed client-side using WebAssembly and WebGL for rendering, leading to near-instantaneous results. This approach enhances performance and security, as data doesn't necessarily need to leave the user's machine for processing. Quadratic supports real-time multiplayer functionality, enabling teams to collaborate on data analysis seamlessly. This feature is particularly valuable for product analytics, allowing individuals with varying technical skills to work together on dashboards, consolidating insights from multiple sources like Mixpanel and PostgreSQL.

THE QUADRATIC AI ASSISTANT CAPABILITIES

The AI assistant in Quadratic is designed to understand the context of the entire sheet, including tables and code cells, with the ability to access more details as needed. It can perform actions that a user could, such as formatting text, but also more complex tasks that users might not know how to do. The AI operates with guardrails, analyzing the sheet after each change to correct errors and ensuring the loop continues until the user's request is met without excessive action. The development includes fine-tuning models to better handle the unique challenges of spreadsheet analysis, such as 2D positioning and multi-dimensional data.

SMART CONTEXT MANAGEMENT AND FUTURE APPLICATIONS

To manage large datasets efficiently, Quadratic employs a strategy of summarizing the sheet's content for the AI's context by default, rather than dumping entire datasets, which would be costly and inefficient. The AI can then intelligently request more specific data or column information when necessary. While primarily focused on analytics, Quadratic's architecture could support building more application-like experiences, potentially incorporating form controls like sliders for interactive model adjustments. Its core philosophy remains centered on the spreadsheet as the optimal interface for human-computer data understanding, keeping the user in the loop.

THE EVOLVING ROLE OF FORMULAS VERSUS PYTHON

The conversation suggests a future shift where Python will increasingly supersede traditional formulas for complex analysis within spreadsheets. While simple formulas like 'SUM' remain effective, tasks involving data aggregation, conditional analysis, or intricate data transformations become cumbersome and difficult to read with formulas. Python, with its robust libraries like Pandas, offers a more readable and powerful alternative for these complex operations. Although JavaScript has strong visualization libraries, Python is currently favored for its maturity in data manipulation and transformation tooling within Quadratic.

SCALABILITY AND TEAM COLLABORATION

Quadratic's development team, though small (nine people), demonstrates a focus on hiring intrinsically motivated individuals obsessed with solving complex problems. This approach is crucial for a remote team building a sophisticated product. The company's source-available model on GitHub builds trust, especially for larger organizations handling sensitive data, and also allows for self-hosting. This transparency fosters confidence and enables broader adoption for businesses that require robust security and data governance.

Mentioned in This Episode

●Software & Apps

●Companies

●People Referenced

Common Questions

Quadratic is an AI-powered spreadsheet that integrates Python, database connections, and AI for data analysis. Unlike traditional spreadsheets, it offers native support for code, allows seamless integration with databases, and uses AI to bridge the gap for users, making complex data analysis accessible through natural language prompts.

Topics

AI & Machine Learning Technology & Innovation Programming & Software Data Analysis Developer Tools AI Spreadsheets Python Integration Database Connectivity Web Assembly Interactive Charts Collaborative Tools

Mentioned in this video

softwarePostgreSQL

A relational database management system that Quadratic can connect to and query.

companyMixpanel

A product analytics company whose data can be pulled into Quadratic for dashboards.

companyDesible

A company where Allesio is a partner and CTO.

toolGoogle Sheets

A traditional spreadsheet tool that is contrasted with Quadratic, highlighting limitations in features like CSV drag-and-drop, scrolling smoothness, and interactive chart generation.

softwareExcel

A widely used spreadsheet software to which Quadratic is compared. Quadratic is not fully compatible for export to Excel due to its unique features like code and database queries.

companyNeon

A cloud-native serverless Postgres company that Quadratic will be adding easier support for.

toolSuperbase

An open-source Firebase alternative that Quadratic will be adding easier support for.

toolBolt

A platform whose development in web assembly is compared to Quadratic's foundational work, highlighting how prior technical investment enables capitalizing on new trends like AI.

companyFigma

A design tool that is used as an analogy for Quadratic's zoom and navigation capabilities within spreadsheets.

softwareBigQuery

Google's fully managed cloud data warehouse that Quadratic will be adding support for.

toolMySQL

A database system that Quadratic supports for connection and querying.

companySnowflake

A cloud-based data warehousing company whose platform can be connected to by Quadratic.

softwareChatGPT

An AI chatbot that is mentioned in the context of generating charts, but noted as not being responsive or interactive like Quadratic's charts.

softwareAirtable

A data storage and process management tool, discussed in contrast to Quadratic's focus on analytics.

softwarePandas

A Python library for data manipulation and analysis, natively supported by Quadratic which contributes to its effectiveness.

companyAnthropic

A company that developed a financial research agent example, which could potentially use Quadratic for data modeling in spreadsheets.

companyCanva

A design platform that launched a data grid feature, which is discussed in comparison to Quadratic's analytic capabilities.

companyGitHub

The platform where Quadratic is source-available, contributing to trust and community engagement.

Found this useful? Build your knowledge library

Get AI-powered summaries of any YouTube video, podcast, or article in seconds. Save them to your personal pods and access them anytime.

Try Summify free