Frequently asked questions

TabInt is an AI-powered platform for turning documents and data into answers you can trust and trace. Upload a CSV or Excel file, or documents like credit agreements and data-room files, and the platform detects your schema, extracts key terms, flags quality issues, suggests relevant analyses, and lets you query everything in plain language. Every result traces back to its source, so you can see where each number came from.

It's built for analysts, operations teams, consultants, and anyone who regularly works with tabular data: CRM exports, transaction logs, survey results, support tickets, and financial reports. You don't need SQL or Python experience to use it, but power users will appreciate the depth available through plain-language querying and AI-driven column generation. It's especially suited to private markets, where credit agreements, data rooms, and deal files hold data that must be owned, defined, and traceable to its source.

No. The interface is designed so that uploading a file is the hardest step. Schema detection, quality checks, and analysis suggestions happen automatically. When you want to go deeper, just type a question like "Which segments grew fastest last quarter?" and we handle the translation to structured queries behind the scenes.

Yes. You can get started immediately with real-world demo data, no credit card required. Sign up with Google or a magic link and you'll have a workspace loaded with sample datasets so you can explore every feature before uploading your own data.

We support CSV and Excel (.xlsx) files, as well as documents like credit agreements, data-room files, and deal docs. The platform infers column types and formats automatically, and extracts the key terms from documents into structured data.

Yes. Every value carries full lineage back to its source, down to the clause or cell it came from, so any number can be traced and defended. You can also reconstruct what the data showed as of any point in time.

Up to 5 million rows per file. The platform is optimized for throughput on real-world datasets like multi-year transaction logs, full CRM exports, and large survey result sets. Processing happens on the server so your browser stays responsive.

Your data is encrypted in transit and at rest. It lives in your private workspace and is never shared with other users or used for model training. You can delete your data at any time and it will be permanently removed from our servers.

Yes. Any view, query result, or enriched dataset can be exported as CSV, Excel, or JSON. You can also duplicate datasets within your workspace for iterative analysis without affecting the original.

AI categorization reads your data and assigns each row to a category based on a taxonomy you define, or one the system suggests. You provide a description, choose a model and sample size, and the platform returns categories with confidence scores so you can review and refine before applying to the full dataset.

AI enrichment lets you generate entirely new columns from your existing data using a natural-language prompt. For example, you could add an industry classification column to a list of company names, extract sentiment from free-text feedback, or derive a risk score from a combination of fields. You name the column, write the prompt, and the AI fills it in row by row.

The platform currently offers the latest AI models from major providers: Anthropic Claude, Google Gemini and OpenAI GPT. You choose the model per operation, so you can optimize for speed and cost or accuracy depending on the task.

Yes. The AI chat assistant accepts natural-language questions like "Show me customers who haven't ordered in 90 days" or "What's the average revenue by category?". These are translated to structured queries and executed against your actual data, so results are computed, not hallucinated. You can also chain queries, build on previous results, and export answers directly.

Based on your data's structure the platform surfaces concentration analysis, segmentation breakdowns, outlier detection, and distribution insights. These suggestions are ranked by relevance and appear after upload, so you immediately know where to dig in.

Click any column header to see instant statistics tailored to the data type. For numeric columns, you get mean, median, distribution, and a box plot. For text columns, you see top values, frequency bars, and unique counts. For dates, you get range and gap analysis. Missing values are flagged everywhere.

Every column supports type-aware filtering. Dates get presets or custom range pickers, numbers get min/max sliders, booleans get checkboxes, and text columns let you search and select from unique values, with visual distribution bars showing relative frequency. You can stack multiple filters across columns and the row count updates in real time.

Yes. Core computations (aggregations, counts, distributions, sorts, and filters) are deterministic. There is no sampling or approximation. Every analysis shows its logic and you can export the underlying data to verify results yourself. AI-generated content includes confidence scores for transparency.

Yes. All data is encrypted in transit (TLS) and at rest. Each workspace is isolated. We never use customer data for model training, and you can permanently delete your data at any time.

No. Your data is never used for training any AI model, whether ours or any third-party provider's. When AI features are used, your data is sent to the selected model provider solely for the purpose of generating a response, with no retention for training.

Every analysis shows its logic: the query that was run, the filters applied, and the operations performed. There are no black boxes. You can export any result set and independently verify the computations. Your full query history is also preserved in the sidebar for traceability.

Yes. You can sign up and start working with demo data and your own files at no cost. The free tier includes core features like upload, schema detection, filtering, sorting, statistics, and plain-language querying. AI operations consume credits.

AI-powered operations use credits. Before running any operation, you see a cost estimate so there are no surprises. Credits can be purchased in your workspace and your balance is always visible. The cost varies by model and the number of rows processed.

Pro gives you everything in the free tier plus expanded processing limits, priority model access, and additional workspace capabilities. You can upgrade at any time from within your workspace.