Everything people ask us

CORESight is an AI business intelligence tool and conversational analytics platform made by CoreOps.AI. It lets anyone in a company ask questions about the company's data in natural language and get back a checked answer, the right chart and a short explanation of what it means. It works in four steps. First, you connect the systems you already run, such as SAP HANA, Oracle, Snowflake or a spreadsheet. Second, CORESight reads how your tables fit together and learns the words your team uses, so it knows what you mean by revenue or lead time. Third, someone asks a question the way they would ask a colleague. Fourth, CORESight works out which data answers it, writes and runs the query or analysis, checks every figure against that analysis, and writes up the answer. You can open each step to see the code it ran, so nothing about how an answer was reached is hidden.

Yes. CORESight connects to SAP at the database level, which covers SAP HANA, and S/4HANA or ECC data stored on SAP HANA, Oracle or Sybase. Instead of extracting SAP tables into a separate warehouse and modelling them first, you connect the database and CORESight learns its structure: which tables hold sales orders, deliveries, invoices or stock, and how they join. People then ask questions such as 'Which plants missed their dispatch targets last month?' and CORESight writes the query against those tables, runs it and returns the answer with a chart and an explanation. Because SAP table names are famously cryptic, CORESight keeps its own glossary of what each table and column means, so answers use your business words rather than technical codes. Today CORESight does not connect to SAP C4C, SAP CRM or SAP BW. If you run one of those, write to support@coreops.ai and we will tell you what works for your setup.

CORESight connects to the databases and warehouses most enterprises run on: PostgreSQL, MySQL, Microsoft SQL Server, MongoDB, Oracle, SAP HANA, SAP on Oracle or Sybase, Snowflake, Databricks and Google BigQuery. It also connects to Salesforce and TallyPrime, and it works from spreadsheets and files (CSV, Excel, Parquet and XML), Google Sheets and Google Drive when there is no system to connect to. Most databases can be connected live, which means CORESight queries them where they are and nothing has to be exported first. Some sources, such as Salesforce, are copied into a private dataset when you sync them, which keeps answers fast and stays within their usage limits. Which sources you can use depends on your plan: the free plan works with files, Google Sheets and Google Drive, and paid plans add databases and business apps. If a system you rely on is missing, write to support@coreops.ai, because new connectors are added regularly.

Traditional business intelligence tools such as Tableau and Power BI are built around dashboards: an analyst models the data, designs the charts and publishes them, and everyone else reads what was built. That works well for questions someone predicted. CORESight starts from the question instead. Anyone on the team types what they actually want to know, in natural language, and CORESight writes and runs the query, checks the figures and returns the answer with the right chart and an explanation, usually while they wait. There is no data model to build first and no new dashboard to request when the question changes. CORESight also writes standing reports for each team when you connect a system and refreshes them on a schedule, so the regular pack does not depend on an analyst's backlog. Many teams use both: their BI tool for the dashboards they already have, and CORESight for the questions those dashboards cannot answer.

Yes. Security is the first thing most buyers ask about, and CORESight is built for it. Every organisation is kept strictly separate from every other, and every question works in its own sealed space. With a live connection your tables stay in your database: CORESight runs the query there and brings back only the results the question needs. The AI model sees the layout of your tables, a small sample of rows and those results, which is what it needs to explain the answer, and never your connection details, which are encrypted and hidden even from the AI itself. Admins decide which databases, tables and columns each person can use, and logs are stripped of personal data before anything is stored. CORESight has also been independently penetration tested by a CERT-In empanelled firm, with every finding fixed and confirmed closed on re-test. The full details, including how AI models handle your data, are on the Security and Compliance page.

Yes, there is a free plan, and you can start in minutes. Sign up, upload a spreadsheet or connect Google Sheets or Google Drive, and ask your first question in natural language. The free plan is for one person and one dataset, which is enough to see how CORESight answers on your own numbers. Every plan includes secure data connections and AI-powered analysis. When you want more, pick a paid plan: Starter, Pro and Premium add more people and datasets, connections to databases such as SAP HANA, Oracle and Snowflake, and features such as standing reports, Deep Insights and forecasting. Plans are billed monthly or yearly, with a saving on yearly billing, and prices are shown before 18% GST in the pricing section of the home page. If you would like help choosing a plan or connecting your first system, write to support@coreops.ai and the team will help you set it up.

Yes, that is the point of CORESight. People ask questions the way they would ask a colleague, for example 'Which customers bought less this quarter than last?', and CORESight turns the question into the query or analysis that answers it. Nobody needs to know SQL, the names of your tables or how they join, because CORESight has already learned that from your data and keeps a glossary of your business terms. Questions can be as simple as last month's sales by region or as involved as why margin fell, and the same plain words work for both. Follow-up questions work the same way: ask 'and by region?' and it knows what you meant. People can type their question or speak it. For anyone who does want to check the working, every step of an answer can be opened to show the code that ran and what it returned, so analysts can trust the numbers and business users never have to look at a line of code.

Natural language to SQL, often called text-to-SQL, turns a question written in everyday language into a SQL query that a database can run. SQL is the language most business databases use to store and fetch data. CORESight uses it to answer questions straight from your systems. When a source is connected live, CORESight works out which tables and columns the question refers to, writes the SQL, checks that the query only reads data and never changes it, and runs it on your database. When the data is a file or a stored copy, it runs the same kind of analysis in Python instead. Either way, the number in the answer comes from the query that ran, and you can open the step to see it. For example, 'What were our top ten products by margin in March?' becomes a query that filters March sales, works out the margin for each product and sorts the result.

Yes. Every answer comes with the chart that suits the data, chosen automatically from nine chart types: a line for a trend over time, bars to compare categories, and other types where they fit better. Because live connections query your systems directly, the numbers are as current as your database, which makes it real-time business intelligence rather than a snapshot from last night's extract. Beyond single answers, CORESight writes standing reports for each team when you connect a system, such as margin and cash for finance or pipeline and retention for sales, and refreshes them on a schedule. You can save any answer as an insight, pin charts to your own dashboards, and export results as CSV, or as PowerPoint or PDF on paid plans. When an answer is a single number or a very detailed table, CORESight shows it as a table, and you can ask for the chart you want.

Conversational analytics means getting answers from data by asking questions in plain language, in a back-and-forth conversation, instead of building queries or reading dashboards. It matters for enterprises because most decisions wait on data that only a few analysts can reach. When a finance lead, a plant manager or a sales head can ask 'Why did margin drop in the north last quarter?' and get a checked answer in minutes, the analyst backlog shrinks and decisions move faster. Analysts spend less time on one-off requests and more on the questions that really need them. Follow-up questions keep their context, so a manager can dig from a headline number down to the region, product or customer behind it without starting again. CORESight is a conversational analytics platform built for this: it connects to enterprise systems such as SAP, Oracle and Snowflake, checks every figure against the analysis behind it, and keeps each organisation's data separate from every other.

Generative AI does three jobs in CORESight. It understands the question, including the business words your team uses, and works out which data can answer it. It writes the query or analysis code that produces the numbers, so the figures come from your data and not from the model's memory. And it explains the result in plain language, with what it means and what to look at next. CORESight then checks the numbers in the explanation against the analysis that produced them, so the written answer stays tied to real figures. Every step is visible, so you can see exactly what was done. The same technology reads a newly connected dataset to learn how its tables fit together and to draft standing reports for each team. You can run CORESight on the AI model you prefer, including Claude, Gemini and GPT-class models, with your own API key or ours, and it carries on if one provider slows down.

CORESight gives decision makers answers they can act on, not just tables. A typical answer has the headline numbers, the right chart, a short explanation of what is driving the result and a recommendation for what to look at or do next. It covers descriptive questions such as sales by region, margin by product or overdue receivables, diagnostic questions such as why a number moved, and forward-looking ones such as a forecast of next quarter's demand. Standing reports give each function its own pack: finance sees margin and cash, sales sees pipeline and retention, operations sees stock and lead time. For bigger questions, Deep Insights works through a topic in several steps and writes up the findings while you get on with your day. Admins control who can see which data, so each person gets insights only from the tables they are allowed to use. Every answer can be saved, shared or exported, so the insight reaches the meeting where the decision is made.

CORESight works from your actual data, so the numbers are real. It writes and runs the queries and code that extract exact answers, and checks each figure in the answer against them. Accuracy depends on the quality of your schema metadata (table/column descriptions). Admins can improve accuracy by editing these descriptions.

Predictions are statistical models based on your historical data. They include confidence intervals and should be used as directional guidance, not exact guarantees. The more historical data available, the more reliable the predictions.

Simple queries (aggregation, filtering): 5-15 seconds. Complex analytics (predictions, multi-step): 30-120 seconds. Data Science mode (ML models): up to 5 minutes. You can watch progress in real-time.

Yes. CORESight uses JWT authentication, encrypted database credentials, role-based access control, and multi-tenant isolation. Your data is never shared with other organizations. All queries are scoped to your client ID.

Yes. CORESight keeps the context of the conversation, so follow-ups like "Now show me just the North region" or "Predict for next month" just work. Start a new chat when you move to a different topic.

Click the "Stop" button that appears during streaming. The response will stop and you can ask a new question.

Yes. Click the microphone icon in the chat input. Allow browser microphone access when prompted. Speak your question clearly and the transcribed text will appear.

CORESight notices the error and retries with corrected code. If it still cannot get there, you'll see a clear error message. Try rephrasing your question or being more specific.

Yes. Expand any step in the response to see the Python code, execution output, and timing. This is useful for verification and learning.

Charts are generated automatically when the data is suitable for visualization. Single-value results or very complex multi-dimensional data may only show as tables. Try rephrasing to ask for a trend or comparison to get a chart.

Yes. Tables can be exported as CSV on every plan, and on paid plans answers and reports can also be exported as PowerPoint or PDF.

Bold numbers are key metrics that CORESight highlights for quick scanning. These are the most important figures in the analysis, such as revenue totals, percentages and growth rates.

The Note section contains caveats about data quality, completeness, or assumptions made during analysis. Always read this section for important context.

Yes. Click the Settings icon (gear) in the header. Toggle sections on/off: Summary, Key Metrics, Insights, Recommendations, Follow-ups, Notes.

PostgreSQL, MySQL, Microsoft SQL Server, MongoDB, Oracle, SAP HANA, SAP on Oracle or Sybase, Snowflake, Databricks and BigQuery, plus Salesforce, TallyPrime, Google Sheets, Google Drive and file uploads (CSV, Excel, Parquet, XML). Which ones you can use depends on your plan.

Yes. The number of datasets depends on your plan: by default 1 on the free plan, 3 on Starter, 5 on Pro and 10 on Premium.

Ask your admin to re-run the Schema Explorer. This regenerates metadata for new/changed tables and columns.

Yes. Admins can upload CSV, Excel, Parquet, or XML files in the Admin Dashboard > Database Configuration > File Upload. Multiple files can be uploaded to create a multi-table dataset.

Database credentials are encrypted at rest under authenticated encryption with a server-side key, so tampering is detected as well as disclosure prevented. They are never exposed to end users, never returned in full by the API, and are masked from the code the AI generates.

Go to Admin Dashboard > Users > Add User. User limits depend on your subscription plan (by default Free: 1, Starter: 2, Pro: 5, Premium: 10).

Yes. In Admin > Users > Edit, toggle Table Access for each user. Users can only query tables they have access to, and CORESight declines a question that needs a table the user cannot see.

Click "Forgot Password?" on the login page and follow the instructions. Alternatively, ask your admin (or super admin) to reset your password.

Go to Admin Dashboard > LLM Configuration. Add a provider for the model you want (Claude, Gemini, GPT-class models, Llama, Mistral, DeepSeek and more), enter your API key, and set it as default.

Data Analyst handles standard queries (aggregation, filtering, grouping) quickly. Data Science handles complex analytics (predictions, ML models, statistical analysis) with iterative multi-step execution. CORESight chooses automatically based on your question.

Check schema descriptions (Admin > Database), re-run Explorer if schema changed, rephrase your question more specifically, and provide negative feedback so the system learns.

Use a modern browser (Chrome, Firefox, Edge). Try refreshing the page. If the data is too large or unusual, the chart may not render but the table should still be visible.

Sessions last 24 hours. Simply log in again; your chat history is preserved.

Wait 30-60 seconds and try again. If frequent, ask your admin about upgrading your plan.

Be explicit: say "for the same products" instead of "for them". If the conversation has moved across several topics, start a new chat for the next one.