QueryVeil is a browser-based AI data analyst that runs SQL locally on your files. Free plan available; paid plans from $7.50/month billed annually.
Researched and fact-checked by Bartu Cavusoglu · Updated October 26, 2026
QueryVeil is an AI data analyst that runs in the browser. Users drop in a spreadsheet or data file, ask questions in plain English, and receive answers backed by real SQL queries run on their own data. The company positions it against general-purpose AI tools that read a file and generate text without querying it.
The product is built around a privacy claim: files are loaded into a SQL engine inside the browser and never uploaded to a server. QueryVeil describes this as architecture rather than policy, and says anyone can confirm it by watching network traffic in DevTools.
It is aimed at people who work with data but may not want a notebook setup, including analysts, consultants, finance and operations staff, and general spreadsheet users.
The workflow starts with a file. CSV, Excel, JSON or Parquet data loads into DuckDB WebAssembly, which the site says takes under two seconds for most files. QueryVeil then shows a data profile with statistics, distributions and outlier flags before any question is asked.
From there, the AI turns plain-English questions into SQL, runs them on the actual data, and shows both the results and the queries. Users can inspect, modify or re-run any query themselves, or write their own SQL in the editor.
For bigger questions, Deep Analysis plans a multi-step investigation, runs queries in parallel, follows anomalies, and records findings. The AI then assembles a report that can be edited and presented.
QueryVeil bundles six connected modes in one workspace: Chat, SQL Editor, Visual Explorer, Notebooks, Reports and Deep Analysis. The Visual Explorer lets users view column profiles and data quality and drill into any column without writing SQL, and guided walkthroughs on sample datasets teach the workflow.
Deep Analysis runs up to three steps at once, and Follow the Thread adds steps automatically when it spots an anomaly. Reports use 14 chart types, side-by-side layouts, insight cards and callout blocks, and a Notion-style editor for dragging blocks and resizing columns.
Reports can be presented as slides with smart grouping and keyboard navigation, or exported to CSV, HTML and PDF on paid plans. Data can also come from URLs or from PostgreSQL, MySQL and BigQuery connections.
QueryVeil is a browser-based AI data analyst. It loads CSV, Excel, JSON and Parquet files into a local SQL engine (DuckDB WebAssembly), answers questions with real SQL, and builds reports from multi-step investigations.
The Free plan is $0 forever. Pro is $90 per year ($7.50 per month billed annually) and Premium is $190 per year ($15.83 per month billed annually). A Teams plan is listed as coming soon, starting at $49 per user per month.
Premium includes a 14-day free trial with no credit card required. Pro requires payment upfront. After a Premium trial ends, the account drops to the Free plan and data stays in the browser.
No. Files are processed in the browser and never uploaded. In standard chat, the AI sees schema metadata and the question; in Agent Mode and Deep Analysis it also sees query results capped at 100 rows. Local AI sends nothing out.
It supports CSV, Excel (.xlsx and .xls), JSON and Parquet files, plus URL imports. Premium adds up to 5 connections to PostgreSQL, MySQL and BigQuery. File size is limited by browser memory.
According to the company, ChatGPT and Claude generate text about a file without running real SQL, so answers can be hallucinated. QueryVeil runs actual queries in DuckDB locally and shows the SQL behind each answer.
Local AI runs through WebLLM or Ollama. Paid plans offer cloud AI across 14 models via OpenRouter, including GPT-5, Claude, Gemini 3 and DeepSeek, with 2.5M tokens per month on Pro and 25M on Premium.
QueryVeil says its architecture is designed for GDPR alignment, since files are processed locally and no personal data is sent to its servers. It does not claim a formal certification in the provided sources.