Frequently Asked Questions
Clear answers about Querri: what it is, how the Library works, how it checks its answers, and how it compares to the tools you already use.
Getting Started
What Querri is, who it's for, what it costs and how to get up and running.
Querri is the question-first data platform. You ask questions about your business in plain English, and Querri builds the data, definitions and logic each question needs, then keeps learning your business.
Most data tools start with tables and infrastructure. Querri starts with the questions your business needs answered. It connects to your apps, databases and spreadsheets, organizes everything in the Library around those questions, and answers with tables, charts and dashboards. Because every answer reads from the same definitions, every person and every AI assistant, including Claude and ChatGPT, works from one trusted understanding of your business.
Querri is for companies that want answers about their business without first building a data team or a data stack, and for the people inside them who ask the questions.
It fits three roles. Data consumers are anyone asking questions for their own job, like a sales lead before a pipeline review or a CS manager before a QBR. Data owners know what the data means, often an ops, RevOps or finance lead, and keep its definitions right. The data team or IT owns systems, access and security. Querri is used in SaaS, eCommerce, logistics, senior living, finance, nonprofits and agencies.
Related questions
Sign up free at app.querri.com, tell Querri the questions that matter most to your business, then upload a file or connect a source and start asking.
Setup starts with the question "What keeps you up at night?". The Librarian suggests business questions to keep, and the Library holds them before any data is connected. Once data arrives, Querri takes a first look and works through up to six of your questions. Teams that want a head start can install a Blueprint for their industry, with questions, views, KPIs and dashboards built in, and try it on example data first.
No. If you can describe what you want in plain English, you can use Querri. You don't need SQL, Python, data modeling or a BI tool.
Querri writes the queries and code, runs them against your data and explains the result in plain language. For anyone who wants to check the work, every answer shows the query behind it.
For example, type "Compare this quarter's sales to last quarter by region" and Querri loads the right data, makes the comparison and builds the chart, without you writing a formula.
Anything about your business data, in plain English: quick lookups, comparisons, trends, forecasts, and multi-step analysis that joins several sources.
Quick questions go to the Librarian, which answers from your Library and comes back in seconds when your team has asked something similar before. Bigger analysis runs as a project, step by step, with forecasting, statistics, financial statements and text classification built in. You can ask follow-up questions to drill deeper.
Examples: "Who were our top 10 customers by revenue last quarter?", "Which carriers missed their on-time delivery SLA this month?", "Which accounts haven't ordered in 90 days, by region?", "Forecast next quarter's revenue from the last 24 months."
Related questions
Querri has a free plan with no credit card required. Paid plans (Pro, Team and Scale) are priced on usage, measured in credits, rather than per seat.
A credit is one meaningful interaction with the AI, like asking a question. Viewing dashboards, browsing the Library, uploading files, exporting results and inviting colleagues don't use credits, so people who only read can join without adding licenses. Admins can see every charge on the Usage page, set monthly credit limits per person and get alerts before the plan runs out.
Related questions
Start with the help center at docs.querri.com and the video library. For anything else, contact the Querri team directly.
The help center covers setup, the Library, KPIs, facts and rules, connectors and checking an answer. If a connection is giving you trouble, the Connector Agent can draft a support request with the connector's diagnostics attached and your credentials stripped out.
Related questions
The Library & Librarian
How Querri organizes your data around your business questions, and how you search it and teach it.
The Library is your company's single source of truth for what its numbers mean. It organizes your data around your business questions, each tied to the KPI that measures it and the view that answers it.
Each collection gathers one topic around a main question and its follow-up questions, with the views, facts and sources that answer them. A Coverage bar shows how many questions are answered and which are still waiting. Because dashboards, chat and AI assistants like Claude all read from the Library, the board deck and the team dashboard show the same number. Workspaces and roles control who can change each part.
Type in Find anything… to search the Library by meaning rather than exact words, or just ask the Librarian a question in plain English.
Search matches what you mean, so "revenue" also finds sales, bookings and the KPIs and views behind them. Switch from List to Visualize to see the Library as a map, with files and connectors on the outside, collections in the middle, and KPIs and facts at the center. Zoom to highlights everything related to a topic, and while the Librarian works, the items it reads for an answer light up on the map.
For example, a CTO who asks "where does our churn number come from?" can search "churn", zoom to it, and follow the line from the KPI back to the CRM table it's measured from.
Related questions
The Librarian is Querri's AI assistant for asking questions and growing the Library. It answers from the views and KPIs your company already agreed on, and shows how it got each answer.
When a saved view or KPI clearly owns a question, the Librarian reads from it, which is why repeat questions come back in seconds. Otherwise it writes SQL against your source tables with your rules applied. Each answer shows its route (Read from, Queried or Searched your library), the rows it returned and Show the query. To add a new question, collection or view, the Librarian shows its plan on a review card and waits for your OK.
A view is a saved, pre-calculated table built to answer a business question directly, such as revenue by month by channel, with the SQL behind it one click away.
Views are where much of what Querri learns ends up: a rule worked out once becomes a column instead of being re-derived on every question. When a source syncs, every view built on it rebuilds in order, and the KPIs measured from those views recalculate. You can open How it's computed on any view to read its SQL.
Related questions
Each KPI is defined once, with the table it's measured from, a period and the SQL that measures it. When anyone asks for it, Querri quotes the stored value instead of calculating it again.
That's why the number in chat, on a dashboard and in Claude is the same number. Every KPI also maps to at least one of the Five Dimensions of Business Value (Revenue, Cost, Efficiency, Quality and Risk), so you can see when a team tracks revenue but nothing about cost or risk. Adding a KPI takes one sentence, like "Track net retention as a KPI."
Related questions
Tell the Librarian in one sentence, like "sales means paid orders only." Querri saves it under Facts & Rules, and every query and view built after that applies it.
Rules attach to the data they're about and are found by meaning. You can also add them by accepting an Inbox suggestion, attaching a fact to a column, or uploading a document. Upload a policy PDF, SOP or contract and Querri lists the statements it makes, each with its page number, so you can choose which become rules. Upload a data dictionary and answers show "Public, 4-year" instead of a code like 1.
For example, a RevOps lead says "Closed Won means Won" once, and every pipeline number after that counts deals the same way, whoever asks and wherever they ask.
Related questions
Yes. Querri remembers your questions, definitions, rules and KPIs, and each night it reviews your Library and suggests improvements. Nothing changes without your OK.
The nightly pass reviews existing answers, finds questions your data can answer that nobody has asked, spots the same customer in two systems, and checks numbers your documents state against the real data. Its findings arrive as cards in the Inbox, where one click accepts a suggestion. Querri learns your business, but it never trains any AI model on your data.
Data & Integrations
What data sources Querri connects to, how it handles your files, and how it keeps data fresh.
Querri connects to 125+ business apps, databases and warehouses, and reads CSV, TSV, Excel, JSON and Parquet files up to 2 GB each. It also reads PDFs and Word documents for the rules they contain.
Connectors include QuickBooks Online, HubSpot, Salesforce, NetSuite, Shopify, Stripe, Google Drive, Google Sheets and Google Analytics, plus PostgreSQL, MySQL, SQL Server, Snowflake, Redshift, BigQuery, Databricks and MongoDB. Around 90 more SaaS tools connect with an API key. A system without a connector can still be included by dropping in its CSV, JSON or Parquet export. Uploading files doesn't use credits.
Drag and drop files anywhere in the app, or ask the Connector Agent to connect a system for you, like "connect HubSpot and our Postgres."
The Connector Agent plans each connection, connects and tests it, and helps you choose tables. Passwords and API keys go into a secure prompt, so the agent never sees them, and every connection and data load waits for your Approve. Ask it "what's connected, and is anything broken?" for a quick health check. Databases on a private network can connect over VPN.
Yes. Querri joins data from different apps, databases and files in one answer, and works out how a new table links to the data you already have.
When new data arrives, Querri measures each table, maps how it relates to your existing sources and checks it against the questions your business already asks. Where the new data can answer one, a suggestion to build a view appears in the Inbox. The nightly pass also spots when the same customer appears in two systems.
For example, connect Salesforce and upload a spreadsheet of marketing spend, then ask "Which campaigns drove the most closed deals?" Querri matches the two and answers.
Related questions
Querri checks every table as it arrives, with more than 180 checks for over 50 kinds of problem, and helps you fix what it finds before the problems turn into wrong numbers.
The Data Audit lists every issue, worst first, and says which KPI, question or project depends on that column. Typical issues include missing values in disguise like N/A or 1900-01-01, numbers stored as text, spelling variants, mixed currencies and duplicate IDs that disagree. The Janitor then cleans the data with you: it asks only the decisions you have to make, like day-first or month-first dates, and previews each fix.
Related questions
As often as you choose. Each connection syncs on a schedule, either a full refresh daily, weekly or monthly, or an incremental sync hourly, daily or weekly that pulls only new or changed rows.
When a source syncs, every view built on it rebuilds in order and the KPIs measured from those views recalculate, so dashboards and answers reflect the latest data without anyone running an export. If a schedule fails three times in a row, it turns itself off and tells you why, so a broken sign-in doesn't keep failing quietly.
Related questions
By default Querri syncs a copy of each source into your organization's own isolated storage and answers from that copy. On select plans, Snowflake and Databricks tables can stay live in your warehouse instead.
With a live table, rows are never copied. Each question that needs the table sends a read-only query to your warehouse and gets back only the result. You choose live or synced one table at a time, and answers show a Live badge for tables queried in the warehouse.
Dashboards, Reports & Automation
How Querri turns answers into dashboards, scheduled reports, alerts and reusable methods.
Yes. Querri builds the chart each question needs and turns analysis into dashboards you edit by asking, like "add a KPI card for churn" or "use our brand colors."
Charts are written as code, so maps, funnels, waterfalls and animated bar races are one request away, and Querri checks each rendered chart before you see it. To move off another BI tool, paste a screenshot or a photo of a sketch of your current dashboard, and Querri rebuilds it on your own data, calculations included. Dashboard viewers can ask questions about what they see, and Present mode puts a live dashboard on a meeting room screen.
Yes. Schedule any project to run hourly, daily, weekly or monthly, and Querri refreshes the data, reruns the analysis and emails the results.
A Deliver step sends real email with values from your data in the subject and body and files attached, such as an Excel workbook or a chart. A Loop repeats the same steps for every row of a table, so one request builds a report per region, store or rep. Querri previews the first five and sends nothing until you approve. Scheduled runs replay the stored code rather than asking the AI again.
For example, a weekly carrier performance report can pull fresh shipment data every Monday at 7 a.m., rebuild the charts and email each regional manager their own version.
Yes. Ask the Librarian to "keep an eye on revenue by store," and a Monitor watches each store against its own history and tells you which one moved, not just that the total did.
Monitors check every time the watched data rebuilds and look for unusual readings, lasting shifts, gradual trends and changes in direction. You control how often they speak up with a slider from Fewer alerts to More alerts, not a threshold. Alerts go to your Inbox and by email to up to 25 people. Each night, Querri also suggests KPIs worth watching.
Related questions
A Skill saves your team's method for an analysis, so the AI follows the way you do it instead of inventing a new approach each time.
Select the steps of an analysis that worked and click Save as Skill, and Querri drafts the title, description, instructions and example plan. You can also write one from scratch. In chat, Querri loads the right Skill on its own when the question matches. An admin can share a Skill with the whole organization, and Skills export as .qskill files to reuse elsewhere.
Related questions
Yes. Querri builds formatted Excel workbooks with live formulas, and Wrapped turns an analysis into a branded presentation you can download as PowerPoint or PDF.
An Excel export can include a sheet per step, bold headers, frozen rows, currency formats, data bars and color scales, plus formulas like SUM and VLOOKUP across sheets. Querri checks the workbook and fixes any formula that errors before you get the file. Wrapped usually produces 8 to 12 slides. Tables also download as CSV or Parquet and charts as PNG or SVG.
Related questions
AI Assistants & API
Using Querri from Claude, ChatGPT and other AI tools, through the API, or inside your own product.
Yes. Querri connects to Claude, ChatGPT, Gemini Enterprise, Cursor, VS Code and other MCP clients, so your AI assistant answers from your company's definitions instead of guessing.
The assistant signs in as you and sends questions to the same Librarian the app uses, so it follows your Library's KPIs and rules and only sees data you're allowed to see. If a question is ambiguous, it asks a short clarifying question rather than guess. Dashboards and charts come back as interactive cards in Claude, ChatGPT and VS Code. Admins can add Querri for a whole Claude Team or Enterprise organization in one step.
Yes. Querri has a public REST API, a Python SDK and a command-line tool, all working under the same access rules as the app.
The API covers projects and chats with streamed answers, sources, views, dashboards, the Library, sharing, users, row-level access policies and embed sessions. Each API key carries its own scopes, rate limit, expiry and optional IP allowlist. Install the SDK with pip install querri, or add the CLI with pip install 'querri[cli]'.
Yes. You can embed Querri's dashboards, Library and chat in your app with an SDK, or run Querri as a fully white-labeled product under your own brand and domain.
The embed SDK has components for React, Vue, Svelte and Angular, plus PHP and WordPress support, and each of your customers sees only their own data through row filters you set. White-label partners get their own tenant with their own database, brand theme, logo, product name and domain. Querri runs in English, Spanish, French, German, Portuguese, Arabic and Hindi.
Related questions
Accuracy & Trust
How Querri checks its work, shows where every number comes from and says when it can't answer.
Yes, and you can check. Querri tells you when a question has more than one fair reading, works answers out two independent ways, and shows the query behind every number.
Before it writes a query, Querri checks the columns involved for gaps, duplicates and values that won't parse. It then works the answer out a second way and tells you if the two disagree, and checks every number in its written summary against the actual result. The AI never does the math itself: it writes SQL and Python that run against your data. On ambiguous benchmark questions, Querri was right 97% of the time measured against the source data, versus 80% for a leading general-purpose AI agent.
Click Show the query to read the SQL behind a Librarian answer, or click any number in a project's answer to open Where this number comes from.
Where this number comes from explains the figure in one sentence, lists every step it depends on (joined, filtered, grouped, summed), and shows each source table with its connector and when it was last refreshed. The explanation is read from the code that actually ran, not written by the AI. The same panel explains any dashboard tile.
Related questions
It tells you. If the data can't answer the question, Querri says so and explains what's missing instead of making up a number.
In Librarian chat this shows as Query declined, with no number attached. When a question has two reasonable readings, like average deliveries per day worked versus per calendar day, Querri says which one it used and shows how the other would change the answer.
Related questions
Yes, for the same data. Re-running a saved view, project, dashboard or scheduled report replays the code that ran the first time, and no AI model is asked to redo the math.
A new question is worked out fresh from the same definitions and rules. Only steps that use AI on purpose, such as text classification or a Wrapped presentation, call a model again on each run. This is why an auditor re-running last month's commission report sees the figure they saw before.
Security & Privacy
How Querri protects your data, controls access and keeps your analysis private.
Yes. Querri is SOC 2 Type II, HIPAA and ISO 27001:2022 certified.
Each customer's data lives in its own database and storage, encrypted at rest with AES-256 and in transit. The AI runs on Amazon Bedrock and Microsoft Azure in the US. The Audit Log records every access-policy change, every API key created, revoked and used, and every data read through an API key, with who, when and the IP address. Reports and policies are available in the Trust Center.
Related questions
Only the people you give access to. Roles control who can view or edit each workspace and item, and row-level access policies control which rows each person sees.
An admin writes an access rule in plain terms, like "store_id is one of 101, 102," and assigns it to people or API keys. The rule applies everywhere: chat, the Librarian, dashboards, Claude and ChatGPT, the API and embeds. It's applied to the data before the AI's query runs, so no prompt can talk its way past it. A shared dashboard re-runs for each viewer, so everyone sees only their own rows.
No. To answer a question, the AI sees only small samples of your data. It writes code, and that code runs inside Querri against the full data with your access rules applied first.
This keeps your full data inside your organization's isolated storage. It's also why answers are calculated rather than estimated: the numbers come from running real queries, not from the model reading and summarizing rows. The Connector Agent never sees passwords or API keys either.
Comparisons
How Querri compares to other tools you may be evaluating.
ChatGPT and Claude are general-purpose AI assistants. Querri is a data platform that gives them, and your whole team, one agreed understanding of your business data, and you can use it from inside either one.
Pointed at raw exports, a general AI assistant has to guess what your numbers mean, so the same question can get different answers from different people. Querri keeps your data connected and current, saves your definitions and rules in the Library, shows the query behind every number, and turns answers into dashboards and scheduled reports. Connect Querri to Claude or ChatGPT and the assistant answers from those same definitions.
Related questions
Tableau and Power BI need clean, modeled data and someone to build each dashboard. Querri covers the whole path from raw data to dashboard, and you build and change it by asking in plain English.
With a traditional BI tool, a data engineer typically builds the pipeline, the semantic model and the measures before a business user can make a chart. In Querri, the Library holds your definitions, the Librarian builds the views, and dashboards are edited with a sentence. To move an existing report, paste a screenshot of it and Querri rebuilds it on your data.
ETL tools move and model data for data engineers. Querri connects, cleans and models data too, but its goal is the answer: it takes your business question all the way to a number, chart or report.
Querri includes 125+ connectors, sync schedules, data cleaning and views that rebuild in order when sources change, so many teams don't need a separate pipeline to get started. If you already run Fivetran, dbt and a warehouse, Querri connects to the warehouse as one more source and gives business users a way to ask questions of the modeled data.
Querying a warehouse directly takes SQL and knowledge of the schema. With Querri, you ask the same question in plain English and get the answer, along with the SQL that produced it.
Querri connects to Snowflake, BigQuery, Redshift, Databricks and other warehouses, either syncing tables or, for Snowflake and Databricks on select plans, querying them live. Your definitions and rules are applied automatically, so business users get the same numbers your analysts would. If you don't have a warehouse, Querri gives you a place to bring your data together without building one.
A question-first data platform starts from the questions a business needs answered and builds the data, definitions and logic each question needs, instead of starting with tables and infrastructure.
In a traditional data stack, one product moves the data, another stores it, another defines it, another charts it, and the AI sits on top. Meaning gets lost at each handoff, and the business question is designed for last. Querri does all of it in one product, organized around your questions, so every person and every AI agent works from the same understanding.
Related questions
Large language models are AI systems that understand and generate language. Querri uses them to turn your plain-English questions into SQL and Python that run against your data, never to guess the numbers.
The model plans the analysis and writes the code. The code runs inside Querri with your access rules applied, and the results, charts and summaries come from that code. Querri then checks every number in the written summary against the result. Saved reports replay the stored code, so no model is involved when they re-run. Paid plans use a larger reasoning model for project analysis.
Related questions