QQuadratic

Quadratic is a browser spreadsheet whose AI agent writes and runs Python, SQL, JavaScript and formulas. It can read a database schema, write the query, clean the result in Python and lay it out across sheets, using 19 named tools. Pro costs $18 per user a month on yearly billing with $20 of monthly AI credits, and a Personal plan is free.

It suits analysts and finance or operations teams who are comfortable reading code and want their data, queries and charts in one shared file. Teams that only use simple formulas may find it more than they need. We rate it 9.0/10 on our scale, with high marks for pricing, integrations and early user feedback.

Quadratic by the Numbers

Quadratic publishes its prices and product changes on its own site, and its seed announcement names investors and angels.

The investor list says a lot about the audience. Wes McKinney created Pandas, the Python library most analysts use for tabular data, and Quadratic puts Python inside the spreadsheet grid. That is a bet on analysts who code, which the agent’s SQL and Python tools reflect.

The credit numbers set expectations for cost. Pro’s $20 of AI credits a month is a fixed budget per user, and Business doubles it with the option to buy more. Heavy daily agent use will show up in credit top-ups, not in seat prices.

The technology row explains why Quadratic can hold large datasets in a browser tab. Rust compiled to WebAssembly and a WebGL canvas let the grid render big tables, which is why the company pitches it for analysis that would slow down a classic web spreadsheet.

Inside a Quadratic Agent Task

Quadratic mixes a classic grid with code cells. The agent works across both. The walkthrough below follows Quadratic’s changelog post and pricing page.

Connecting Data Sources

A team first adds connections. Quadratic lists Postgres, MySQL and Snowflake for databases, QuickBooks for accounting, Google Analytics and Mixpanel for product and marketing data, and Plaid for bank data. Free Personal accounts have limited connections; Pro and Business remove the cap.

What the Agent Can Do on Its Own

The August 2025 update gave the AI access to nearly every part of the spreadsheet. For code, it can read a code cell’s value, write SQL into a cell and rerun code. Before writing SQL, it can read database schemas so the query uses real table and column names. For structure, it can add, duplicate, rename, move, color and delete sheets, insert rows and columns, resize them, set borders, change table column settings and add validations. It can also search text and check whether cells hold data.

Two Chats for Two Jobs

Quadratic keeps a chat on the left for general requests, which can now run SQL directly, and a code chat on the right tied to a specific code cell. Users can ask the left-hand agent for a whole analysis or open a cell and ask for a change to that code.

A Monthly Revenue Analysis Example

A finance analyst connects Snowflake and QuickBooks. She asks the agent to pull monthly revenue by product for the last 24 months, join it with refunds from QuickBooks, calculate net revenue and growth, and put a chart on a new sheet called Summary. The agent reads the Snowflake schema, writes the SQL into a code cell, writes a Python cell for the join and the growth math, adds the Summary sheet, and formats the table. She opens the SQL to check the date filter, tweaks it, and reruns. The finished file holds the queries, so next month she reruns them.

Cleaning a Messy Export Without a Database

Not every job starts with a database. An operations lead uploads a CSV of 80,000 support tickets with inconsistent dates, duplicate rows and free-text categories. She asks the agent to standardize the dates, drop duplicates, map the free-text categories to six standard ones and build a weekly count by category. The agent writes a Python cell to parse dates and remove duplicates, a second cell to map categories, then adds a sheet with a pivot-style table and a chart. Where the mapping is unclear, she edits the list of category rules in the Python and reruns. Quadratic’s seed post says it imports hundreds of thousands of rows, so files of this size fit the product’s stated range.

Sharing Results With Colleagues Who Do Not Code

The finished file is still a spreadsheet. A sales director who never opens a Python cell can read the summary table, filter it and leave comments, while the analyst keeps the code that produced it. Quadratic Teams shares files among members by default, so a workspace becomes a library of live analyses. When the data source updates, the analyst reruns the code cells and the summary refreshes. That split, code for the builder and a grid for the reader, is the main reason to pick Quadratic over a notebook.

What Analysts Keep Control Of

Analysts still check every query and Python step, since the agent’s work is visible as code. They decide which connections to add, which numbers matter and what leaves the team. The visible code is the main safeguard: anyone can see how a number was built.

Quadratic Plan Prices and AI Credits

Quadratic shows yearly prices on its pricing page and says yearly billing saves 10%.

Yearly Costs for Three Data Teams

• A solo analyst on Pro: $18 a month, or $216 a year, with $240 of AI credits over the year.
• A 6-person analytics team on Pro: $108 a month, or $1,296 a year. Pro bills every user on the team.
• A 6-person team on Business: $216 a month, or $2,592 a year, with $40 of credits per user each month and top-ups available.

Seats and credits behave differently. Seats are fixed per person, and Pro bills every user in the team, so adding a part-time viewer still adds a seat. Credits are a monthly budget attached to each paid plan. A team with one heavy agent user and five light users may get more from Business for the heavy user than from upgrading everyone.

Credits are the variable cost. A user who runs long agent sessions every day may burn through $20 of credits before month end and need Business or extra purchases. Monthly prices are available through a toggle on the pricing page; the page lists the yearly figures.

Quadratic Compared With Rows AI Analyst, Hex and Julius AI

Quadratic sits between a spreadsheet and a notebook, so its rivals come from both worlds.

Quadratic is the pick here for people who want SQL and Python results to stay in a grid that colleagues can read. Rows is cheaper per seat and friendlier for non-coders. Hex suits data teams who already work in notebooks and want AI there, while Julius AI answers one-off questions about a file through chat. A team that wants both a notebook’s power and a spreadsheet’s readability for business colleagues is the core Quadratic buyer. Browse more on our spreadsheet agents and data analysis agents pages.

Teams That Match Quadratic Well

• Data analysts who live in SQL and Python: the agent writes both and leaves the code visible.
• Finance and operations teams with data in Snowflake, Postgres or QuickBooks: direct connections remove export steps.
• Startups replacing ad hoc notebooks and spreadsheets: one file holds the query, the math and the chart.
• Teams that hand reports to business users: results land in a grid that readers can filter without touching code.
• Healthcare or regulated teams: the HIPAA badge and Enterprise self-hosting give a starting point for security review.

Users Who May Prefer a Different Spreadsheet Agent

• Excel-centric finance teams: Claude for Excel or Agent Mode in Excel keep models in Excel.
• Non-technical managers: they may find code cells harder to check than plain formulas.
• Teams needing heavy daily AI use on a fixed budget: credit caps can lead to top-ups.

Quadratic Adoption Considerations

The agent’s power comes with responsibility for the code it writes. A wrong join or a missing filter in SQL can produce plausible numbers that are off. Ask reviewers to read the query and the Python before trusting a result, especially on the first runs against a new database.

Database credentials deserve care. Give Quadratic read-only users scoped to the schemas it needs, which limits the effect of any mistaken query. Keep production write access out of spreadsheet connections entirely, and rotate the credentials when people leave the team. For bank data through Plaid or accounting data through QuickBooks, decide who in the workspace may see those sheets before sharing files widely, since Teams shares files among members by default. Self-hosting on Enterprise is an option for teams that cannot send data to a cloud service.

Moving in from Excel or Google Sheets takes planning. Simple sheets import easily, while models built on Excel-only features or Sheets scripts may need to be rebuilt with Quadratic’s formulas or code. Start with a report that already pulls from a database, since that is where Quadratic adds the most.

Quadratic Funding and Agent Releases

On April 2, 2024, Quadratic announced a $5.6 million seed round led by GV and launched Quadratic Teams, shared workspaces with files shared among members by default. On August 27, 2025, the company described a major agent update that let the AI take many more actions in the sheet, with 19 named tools and the ability to run SQL from the main chat.

How Quadratic Scores as a Spreadsheet Agent

Quadratic scores 9.0/10 on our scale. Pricing value and integrations score high because prices are published and the agent reaches databases directly. Autonomy is good: the agent writes, runs and lays out code-based analysis. Setup scores a little lower because the best results need connections and code-reading skills. User sentiment is high but based on only 3 Product Hunt reviews.

For an analyst who already writes SQL, the free plan is enough to test the agent on a real database within an afternoon. Pick one recurring report, connect its source with a read-only user, ask the agent to rebuild it, and compare the code it writes with your own. If the queries hold up, Pro at $18 a month on yearly billing is a small step for the rest of the team.

Analyst Questions About Quadratic

How Much Does Quadratic Cost?

Personal is free. Pro is $18 and Business $36 per user per month, billed yearly, with $20 and $40 of monthly AI credits. Enterprise is custom.

Can the Quadratic Agent Write Python and SQL?

Yes. Cells support Python, SQL, JavaScript and formulas, and the agent can write code into cells, rerun it and read database schemas before writing SQL.

Which Databases Does Quadratic Connect To?

Quadratic lists Postgres, MySQL and Snowflake, plus QuickBooks, Google Analytics, Mixpanel and Plaid, with more on its connections page.

Is Quadratic Secure for Sensitive Data?

Its site shows SOC 2 and HIPAA badges through Vanta, and Enterprise adds SSO and a self-hosting option.

How Is Quadratic Different From Excel With Copilot?

Quadratic is a separate spreadsheet built around code cells and database connections. Excel with Copilot keeps you in Excel and works mainly with formulas, PivotTables and charts.