Quadratic
Quadratic is an AI spreadsheet that connects live to databases, files and APIs. Ask a question in plain English and it writes formulas, Python, SQL and charts for analysts, data, finance, product and marketing teams.
What is Quadratic?
Quadratic is a web spreadsheet built around an AI assistant and live data connections. It runs in the browser on a Rust engine compiled to WebAssembly with a WebGL canvas interface, described by its publisher as the world's first canvas spreadsheet and designed to run at 60 frames per second. By default, data is stored and calculations are executed on the user's own machine.
Cells accept four languages: Excel-like formulas, Python, SQL and JavaScript. The AI writes and edits that code, inserts data and generates charts, but every cell stays open and editable and a step-by-step explanation is displayed beside the generated code, which makes an analysis auditable rather than opaque.
Data arrives through live connections to Postgres, MySQL, MS SQL, Snowflake, BigQuery, MariaDB, CockroachDB, Supabase and Neon, alongside CSV, Excel, Parquet and PDF files. Business connectors cover QuickBooks (P&L, balance sheet, invoices), Google Analytics, Mixpanel, Salesforce and Plaid, which reaches more than 12,000 financial institutions. A sheet holds up to 100 million rows, a hundred times the row limit of Excel, and accepts multi-gigabyte CSV and Parquet imports.
Outside the browser, Quadratic publishes an official MCP server at mcp.quadratichq.com with revocable OAuth access for ChatGPT, Claude, Cursor, VS Code, Windsurf, Zed, Gemini CLI and Codex CLI, and a public REST API (OpenAPI 3.1) that reads and writes cells, code cells, imports and batches, with official Python and npm clients. Scheduled tasks, PDF data extraction, charts, search and Agent Sync cover automation, while real-time collaboration adds visible cursors, version history, cell comments and pivot tables.
The AI models available are Claude Sonnet 5 (default), Claude Opus 4.8, Claude Haiku 4.5, Gemini 3.5 Flash and Gemini 3.1 Pro. On security, the publisher reports SOC 2 Type II and HIPAA, TLS 1.2+ in transit, AES-256 at rest, role-based access control and execution in isolated containers; Enterprise customers can self-host on Docker, AWS, Azure or GCP and bring their own model. Quadratic Growth Inc. was founded in Boulder, Colorado in 2022, claims more than 250,000 users, and keeps an active changelog, with v0.26.12 released on 11 August 2026.
What it does
- Connect databases, files and APIs into a single spreadsheet
- Ask a question in plain English and let the AI write formulas, Python, SQL and charts
- Open and edit any code cell to check how an analysis was produced
- Work across up to 100 million rows, with instant sorting, filtering and analysis
- Drive your sheets from ChatGPT, Claude or Cursor through the MCP server
- Automate work through the REST API and scheduled tasks
- Share a live, revocable link instead of exporting a file
When to use Quadratic / When not to
A quick filter to help you decide if Quadratic is the right fit.
When to use Quadratic
- Data and BI analysts who want to query Postgres, MySQL, Snowflake or BigQuery from a familiar grid instead of learning a separate BI tool
- Finance, accounting and trading teams pulling QuickBooks data, Plaid bank feeds and market data into a single sheet
- Product, marketing and sales teams that need analysis they can audit, since the AI writes visible, editable formula, Python and SQL cells
- Teams working past ordinary spreadsheet limits, with up to 100 million rows per sheet and multi-gigabyte CSV or Parquet imports
- Regulated and security-conscious organizations that need SOC 2 Type II, HIPAA and admin controls over exports, copying and sharing
When not to use Quadratic
- Individuals expecting an unlimited free tool: the Personal plan caps AI usage, sharing, files and connections
- Teams that require EU data residency: the services are hosted in the United States, with no European region and no Article 27 EU representative appointed
- Mobile-first users: no native iOS or Android application was found
- Non-English-speaking users: the interface and documentation are published in English only, with no localization announced
- Small teams that need self-hosting or a self-service HIPAA BAA: self-hosting is reserved for the Enterprise plan and the BAA goes through a sales demo
How to use Quadratic
A typical end-to-end flow, from setup to results.
- Open app.quadratichq.com and create a free account: there is no environment to set up and no package to install
- Import a file by drag and drop (CSV, Excel, Parquet or PDF), or add a SQL connection using read-only credentials
- Check the connection: the sheet refreshes on its own and shows when the data was last pulled
- Open the AI chat panel from the sparkle icon in the sidebar and ask your question in English
- Let the AI pick and write the cell, whether a formula, Python or SQL, while the context gauge shows how much of the model window is in use
- Open the code cell to check or correct the method; the explanation is displayed alongside it
- Change model from the selector at the bottom of the chat panel, or use voice input, available since v0.26.9
- Share a live link with view or edit rights granted per person and revocable at any time
- Connect an AI client through MCP: one-click install for Cursor and VS Code, a 30-second setup for Claude and ChatGPT, then an OAuth authorization
- For automation, create a token under Team Settings > API Tokens and call the REST API with an Authorization: Bearer header
Pros & Cons
Pros
- The analysis stays readable: the AI produces visible, editable code rather than a black-box answer
- A familiar spreadsheet interface with no separate BI tool to learn, on a Rust and WebAssembly engine that calculates locally by default
- Live connections rather than CSV exports, so the figures refresh
- Volumes out of reach of Excel and Google Sheets, with up to 100 million rows per sheet
- Four languages in the same sheet: formulas, Python, SQL and JavaScript
- Zero-day retention at every model provider and no training on customer data, alongside SOC 2 Type II, HIPAA, TLS 1.2+ and AES-256 encryption and admin-level export controls
- A REST API free on every plan, an official MCP server, Enterprise self-hosting with Bring Your Own Model, and an Education plan that is entirely free for students, teachers and researchers
Cons
- Prices are displayed on annual billing only; reading the monthly rate means switching the selector on the pricing page
- No separate free trial: the pricing FAQ points to the free account instead
- The Personal plan limits AI usage, sharing, files and connections, and the more capable models (Claude Opus 4.8, Claude Haiku 4.5, Gemini) are reserved for paid plans
- AI credits are capped in value, at 20 USD per month on Pro and 40 USD per month on Business, so the real cost depends on usage
- Hosting is in the United States only, with no EU data residency option
- No DPA is published and no public subprocessor list can be verified, the Vanta trust center being unreadable without JavaScript
- No native mobile app, an English-only interface, SSO and self-hosting restricted to Enterprise on quotation, and purchases described as non-refundable
Pricing & Plans
Quadratic follows a freemium model. A free plan exists, called Personal, and it is permanent. The cheapest paid entry point is the Pro plan at 18.00 USD per user, per month, billed annually; Business is 36.00 USD per user, per month, billed annually, and Enterprise pricing is available on quotation only. Billing is per seat and prorated when a member is added, payment is handled by Stripe, and the REST API is included free on every plan. Purchases are described as non-refundable, and a subscription can be cancelled at any time, taking effect at the end of the paid period.
- limited AI usage
- limited sharing
- files and connections
- limited support
- 20 USD in AI credits per month
- access to additional models
- additional privacy controls
- unlimited sharing
- files and connections
- 40 USD in AI credits per month
- on-demand purchase of additional AI credits
- everything in Pro
- custom AI usage
- Single Sign-On (SSO)
- self-hosting option
- dedicated support
- unlimited sheets
- code
- teammates and connections
- with AI usage kept at the free plan's quota
Data, GDPR & hosting
A consolidated view of how Quadratic handles your data.
GDPR overview
The site never explicitly claims GDPR compliance, but its privacy policy (version 14, effective 9 April 2026) addresses the regulation directly. A dedicated section sets out the legal bases relied on under the EU and UK GDPR: consent, performance of a contract, legitimate interests, legal obligations and vital interests. Users in the EEA, the United Kingdom, Switzerland and Canada are granted access and copy, rectification, erasure, restriction, portability, objection, and the right not to be subject to a solely automated decision, with an explanation of the main factors and the option to request human review. Consent can be withdrawn at any time, and a data subject access request form is provided. Complaints may be lodged with a member state supervisory authority, the UK authority or the Swiss federal data protection authority. No Article 27 EU representative and no data protection officer is appointed; privacy requests go to support@quadratichq.com.
Who owns the data?
Under the terms of service, Quadratic Growth Inc. asserts no ownership over user contributions: you keep full ownership of your content and of the intellectual property rights attached to it. The company may access, store, process and use the data you provide, within the limits of its privacy policy (effective 9 April 2026) and of the settings you choose. Two exceptions apply. Submissions, meaning suggestions, feedback and ideas sent to the company, are assigned to it outright, with no compensation. The company may also de-identify and aggregate usage data and publish generalized descriptions of it, excluding identifying data and raw contributions without written consent; users can opt out by emailing support@quadratichq.com.
Reuse rights
The privacy policy, effective 9 April 2026, lists the personal data collected: email addresses, passwords and other contact and authentication details provided at sign-up, plus automatically collected log and usage data, device data and IP-based location data. No sensitive data is processed and no data is collected from third parties. Content submitted to the AI features is passed to named AI service providers, namely Anthropic, OpenAI, Amazon Bedrock and Google Cloud AI, for AI development, natural language processing, AI insights, AI research, AI search and text analysis; every model runs under a zero-day retention policy, so inputs and outputs are neither stored nor used for training. Payment details are held solely by Stripe, Plaid aggregates financial-institution data at the user's own instruction, and Google Analytics is active with remarketing, Display Network impression reporting and demographics and interests reporting. Data may also be shared in a business transfer, with business partners, and with other users in the public areas of the service. Under CCPA the company declares that it does not disclose, sell or share personal information.
Data retention & training
Hosting summary
Section 12 of the terms states that the services are hosted in the United States, and accessing them from another region is treated as consent to transferring data there. The structured privacy record declares no international transfer mechanism. The observed infrastructure is consistent with this: the domain resolves to 76.76.21.21, an anycast node on AS16509, Amazon.com, Inc. AI processing involves further providers, namely Anthropic, OpenAI, Amazon Bedrock and Google Cloud AI. Data is encrypted with TLS 1.2 or above in transit and AES-256 at rest, and code runs in isolated containers with resource limits, separated from other customers. No European hosting region is offered, which matters for buyers subject to data residency requirements: the only way to keep data within a chosen jurisdiction is the Enterprise self-hosting option, available on Docker, AWS, Azure or Google Cloud Platform, where the data never leaves the customer's own infrastructure.
Things to keep in mind
Risks and trade-offs to weigh before adopting Quadratic.
- The AI writes the analysis, and it is tempting to accept a result without opening the code cell: the audit trail only helps when it is actually read
- Prices are shown on annual billing by default, so the monthly rate has to be checked before committing
- AI credits are capped in value, at 20 USD per month on Pro and 40 USD on Business, so the real monthly cost depends on how much the assistant is used
- Purchases are non-refundable under the terms, and subscriptions renew automatically until cancelled
- Hosting is exclusively in the United States and using the service is treated as consent to that transfer; no Article 27 EU representative and no data protection officer is appointed
- No DPA is published, the HIPAA BAA goes through a sales demo, and the Vanta trust center cannot be read without JavaScript, so the certifications cannot be checked automatically
- The terms assign Submissions, meaning suggestions and feedback, to the publisher, and personal data may be retained for up to 72 months after an account is closed
Setup & Integrations
Technical difficulty
Low for everyday use: Quadratic is a web application, a free account is enough, with no environment to configure, no package to install and no versioning to manage; a desktop app is also available. Python, SQL and formula skills help but are not required, since the AI writes the code. Moderate for database connections, which need credentials, and for the MCP server and REST API: one-click install for Cursor and VS Code, a 30-second setup for Claude and ChatGPT, an OAuth grant, and a token from Team Settings. High only for Enterprise self-hosting on Docker, AWS, Azure or GCP.
Deployment
Integrations
Supported languages
Behind Quadratic
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Quadratic.
Frequently asked questions
What is Quadratic?
Is there a free plan?
Is there a free trial?
How much does the first paid plan cost?
Is my data used to train the AI models?
Which AI models are available?
Which data sources can I connect, and how large can a sheet be?
Is there an API, and can I drive Quadratic from ChatGPT or Claude?
Is Quadratic certified, and can I host it myself?
Where is my data hosted, and how long is it kept?
Should you pick Quadratic?
Quadratic positions itself as a spreadsheet rather than a BI tool: the grid, the formulas and the sharing model are familiar, and the AI layer sits on top of them. Its distinguishing choice is transparency, since the assistant writes formulas, Python or SQL into cells that stay open, editable and explained, so an analysis can be checked instead of taken on trust. Around that sit live connections to the main databases and warehouses, sheets holding up to 100 million rows, an official MCP server, a REST API free on every plan, and SOC 2 Type II and HIPAA on the security side, with zero-day retention at every model provider.
The reservations are practical rather than fundamental. Pricing is displayed on annual billing, so the monthly figure takes a click to find, and there is no separate free trial: the free Personal plan plays that role, with capped AI usage, sharing, files and connections. Hosting is in the United States only, with no EU data residency option, no Article 27 representative and no published DPA, which matters for European buyers. There is no native mobile app, and the interface is English-only.
The natural audience is therefore analysts, data and BI teams, finance and trading desks, and product, marketing and research groups that already work in spreadsheets and want database-scale data and auditable, AI-written analysis in the same place. The product looks mature for its age: Quadratic Growth Inc. was founded in Boulder, Colorado in 2022 and the changelog is well fed, with v0.26.12 shipped on 11 August 2026. Teams under strict infrastructure constraints have the Enterprise route, with self-hosting on Docker, AWS, Azure or GCP and Bring Your Own Model, while students, teachers and researchers can use the free Education plan.
- Choosing a selection results in a full page refresh.
- Opens in a new window.