# Ask your data a question, get the dashboard. With the access each person already has. > Staff ask Claude or ChatGPT and get dashboards from your own data, read-only and within their rights. Always up to date, saved in your BI tool and reusable. Source: https://schuetze-web.com/integrations/data-analytics Staff ask their AI about revenue, orders or costs and get a chart built from your own data, read-only and within their rights. Dashboards are saved in your BI tool, stay up to date on their own and can be used again by the whole team. ## What changes ### Answers without waiting A question to the AI instead of a ticket to the BI team or IT. Staff explore the numbers themselves, and the specialists keep their time for the hard questions. ### Always up to date A dashboard is not an export. It queries your data again every time it is opened, so it shows the current state instead of last month’s spreadsheet. ### Built once, used again Every dashboard is saved in your BI tool. Colleagues open it again tomorrow, share it with their team or receive it every Monday by email. ## How it connects ### Your systems ERP, CRM, accounting, shop, production Read through a read-only replica or your data warehouse ### Analytics connection Reads your data with its own read-only user, applies the rights of each person and works with your defined metrics. Every query is logged. ### Your team’s AI - Claude - ChatGPT - Gemini - Mistral Also Microsoft Copilot, or open models such as Llama, Qwen or DeepSeek on your own servers ### Your BI tool Metabase, Power BI, Looker Studio, Superset or the one you already use Dashboards are saved and shared here ## One question, two ways Sales wants revenue per customer group for the quarter. ### Today - Sales writes a ticket to the BI team or IT - It waits in the queue for a few days - Someone builds an export to Excel - Sales builds the chart by hand - A week later the numbers are out of date - Next quarter the same ticket comes in again Days of waiting, a spreadsheet that ages, and different numbers in every meeting. ### With the analytics integration One sentence, a dashboard that stays and updates itself, and the same numbers for everyone. ## Numbers you can trust ### Read-only The AI uses its own database user without write rights, on a replica or warehouse. Your production systems are neither changed nor slowed down. ### One definition per metric Revenue, margin or active customer are defined once. The AI calculates with these definitions, so everyone gets the same numbers. ### Traceable Every chart comes with its query and source tables, and the AI names its assumptions. ### Asks when unclear If a question can be read two ways, the AI asks instead of guessing. ## Dashboards that stay What someone builds in a chat is saved in your BI tool. It updates itself, can be shared with the usual permissions, sent out on a schedule and extended in the next chat. ## What your team asks See where customers stand, without a report request. Receivables and costs at a glance. Spot bottlenecks before they cost money. The figures for the next meeting, current on the day. ## Set up in three steps ### Connect read-only We set up a read-only replica or connect your data warehouse, with its own database user without write rights. ### Define metrics and rights Together we define the key metrics and who may see which data, down to regions and individual columns. ### Roll out to the team Your staff ask their AI, and the dashboards land in your BI tool. ## You stay in control You decide which data the AI may read and who sees what: sales only its own region, personal data such as salaries hidden. Queries have time and size limits, and every one is logged with who asked. - Sales · Own region, customers, orders · Salaries - Finance · Revenue, costs, receivables · Personal data - Management · Everything · Nothing ## Questions ### Which data sources can be connected? Most databases and systems that hold business data: ERP, CRM, accounting, shop and production systems, data warehouses such as BigQuery, Snowflake or PostgreSQL, and spreadsheets. In the intro call we look at yours. ### Can the AI change data? No. It works with its own read-only database user, on a replica or a data warehouse. It cannot write, delete or slow down your production systems. ### How up to date are the numbers? As current as the data the integration reads. A replica is usually seconds to a few minutes behind the live system; a data warehouse depends on how often it is loaded, for example hourly or nightly. ### How do you prevent wrong numbers? Key metrics are defined once and the AI calculates with those definitions. Every chart shows its query and source tables, the AI names its assumptions and asks when a question is ambiguous. ### Can staff see salaries or other sensitive data? Only if they may see them today. Rights apply down to rows and columns: regions, departments or individual fields such as salaries can be hidden per role. ### Do we need a data warehouse? Not to start. A read-only replica of your database answers most questions. A data warehouse is worth it when data from several systems has to be combined regularly. ### Which BI tool do you work with? With the one you already use, for example Power BI, Looker Studio, Superset or Metabase. If you don’t have one yet, we set one up; for many companies Metabase is a good start. ### What does the AI provider see? Only the query results needed for the answer, not your database. Which AI you use and where the data is processed we decide together with you. --- Contact: book an intro call at https://schuetze-web.com/book or write to simon@schuetze-web.com.