What you can connect: structured and unstructured data
The kinds of data Nekt can ingest, from APIs and databases to PDFs and images.
- Understand the difference between structured and unstructured data
- Know which kinds of sources and formats fall into each one
- Discover what you can do with each type of data
Before asking "does Nekt have a connector for tool X?", it is worth taking a step back and looking at the type of data. Almost everything you want to analyze falls into one of two groups: structured data, which is born in rows and columns, and unstructured data, which comes loose in documents, images and text.
The two types of data
Pick a tab: each one explains the type, shows example sources and formats, and what you can do with it.
Data that already comes in rows and columns. Every record has the same fields, the column structure and types are predictable, and you can query it directly with SQL. It is the most common type and what most business tools produce.
Where it usually comes from:
| Category | Example sources or formats | What it brings |
|---|---|---|
| SaaS APIs | HubSpot, Meta Ads, Google Ads, Stripe | business objects (deals, campaigns, payments) in records with defined fields |
| Databases | Postgres, MySQL, SQL Server | transactional tables straight from the schema (orders, users, payments) |
| Tabular files | Excel, CSV, Parquet, Google Sheets | one row per record, named columns, ready to become a table |
What you can do: practically everything you expect from data analysis. You type, clean and join in Silver, model in Gold and consume via BI, Data API or MCP. It is the natural territory of data transformations.
Data that does not come as a table. These are documents, text files and images: rich information, but without ready-made columns. To become analysis, it must first be read, extracted or interpreted, often with the help of AI.
Where it usually comes from:
| Category | Example formats | What it brings |
|---|---|---|
| Documents | PDF, DOCX | contracts, invoices, technical reports, proposals |
| Free text | TXT, Markdown | notes, transcripts, articles, support responses |
| Images | PNG, JPG | photos, screenshots, scanned paper documents |
What you can do: extract fields from a PDF, classify images, summarize or search within text. Unstructured data opens up cases that structured data alone cannot reach, but it depends on a reading step before it becomes queryable information.
Unstructured data in the Catalog is a Growth feature
On the Starter plan you can connect all the structured data you want: APIs, databases and tabular files. Bringing unstructured data into the Catalog (PDFs, images, documents and text for AI to read and process) is a feature of the Growth and Custom plans. If your use case involves processing documents or images, this is what unlocks it.
Learn about the Growth planHow to choose where to start
The quick read: if your data already lives in a spreadsheet, a database or a tool with an API, it is structured and fits right into your plan. If the information that matters is locked in contracts, reports or photos, you are in unstructured territory, and it is worth planning both how to read that content and the Growth plan to bring it into the Catalog. In practice, most projects start with structured data, which delivers value quickly, and move on to unstructured data when a case comes up that only it can solve.
Don't connect a type of data just because you can. Start from the question you want to answer and see which data supports it. A lot of what seems to require documents actually already exists in structured form in some system, and bringing in the table version is usually cheaper and more reliable than interpreting the equivalent PDF.
An accounting firm started by bringing in structured data: the ERP via its database and the month-end closing spreadsheets in Excel. That already answered 80% of the management questions. Then it moved on to unstructured data on the Growth plan, uploading invoice PDFs for AI to extract amounts and dates. The structured data provided a quick foundation, and the unstructured data covered what only existed on paper.
Related material
↗ Guide: Sync types ↗ Go deeper in the docs: connectors in NektUse cases that apply what this guide explains.