Kramata: product and architecture
This Kramata review covers a Windows desktop app for people who clean the same messy CSV or Excel export every week or month. You build the cleanup once as a list of steps — remove duplicates, fix dates, merge in a second file by a matching key, reshape columns — save it as a recipe, and re-run it on the next file with one click. Everything runs on your own PC: there is no account and no cloud service, and files are never uploaded. We based this review on kramata.com, its guides and privacy page, the product's release history on GitHub, and information Kramata supplied to us, which we checked against those public pages on 27 September 2026.
Overview
Kramata is aimed at non-developers: operations, finance and small-business staff who today repeat the same spreadsheet chores by hand, or keep them alive with VLOOKUP formulas that break when a column moves. The product's core idea is the recipe. Each step is written in plain language and the table updates live as you add it, so you can check the result before you save. The saved recipe is a portable .json file, so a colleague can run the same cleanup without rebuilding it.
The product is young. The first public release, version 0.9.0, shipped on 8 June 2026, version 1.0.0 followed four days later, and the current version is 1.0.3, released on 18 September 2026. It runs on Windows 10 and 11 (x64). The Microsoft Store is the vendor's recommended install path, because it installs without security warnings and updates the app automatically; a direct installer is also offered.
Key Features and Architecture
Cleaning steps. Kramata covers the chores most recurring exports need: remove duplicates on any columns, standardise mixed date formats, trim spaces, fix case, fill blanks, find and replace, convert types, filter rows, and split or merge columns. For dates you state whether the file is day-first or month-first, so a value like 04/03 is never guessed.
Merge by key. Merge pulls columns from a second file by matching a key column — the job people use VLOOKUP or XLOOKUP for. It keeps the row count of the main file by taking the first match per key. That is the right behaviour for a lookup, and it avoids surprise duplicate rows. If you need every matching row from the second file, as in a many-to-many join, this is a different operation.
Reshape and calculate. Pivot and group totals rows by sum, count or average. Unpivot folds wide month or category columns into tidy rows in one step. Calculated columns cover totals, differences, percentages and averages across columns without a formula language.
Flagged rows. Rows that fail a check go to a separate Flagged list that you can review and export, instead of being dropped silently. For finance and operations work this is often the step that matters most, because it leaves a record of what the cleanup set aside.
Batch processing (Pro). A recipe can run across a whole folder at once and return each file cleaned separately, or all of them merged into one. The vendor points out that Power Query can stack a folder of files into one table but does not hand back each file cleaned on its own; Kramata's batch mode does both.
Files and engine. Kramata reads CSV, .xlsx, .xlsm and legacy .xls files, and detects columns even when a file has messy multi-row headers. It exports to CSV and Excel. Processing uses DuckDB, an in-process database, so large CSVs load and scroll smoothly, and the vendor states that the same input with the same recipe produces byte-identical output.
Privacy. File names, paths, sheet and column names and cell values stay on the machine. The app sends a small set of anonymous usage events through Aptabase — for example that the app was opened, a loaded file's type with a bucketed row count, the type of a rule added, and error codes — and you can switch this off under Settings & privacy. A Pro licence key is checked once with Gumroad on activation and then stored locally.
Ideal Use Cases
Kramata fits a person who receives the same export on a schedule and does the same cleanup each time: a monthly CRM or ERP export that needs duplicates removed and dates fixed, a supplier price list merged against an internal product list, or a set of branch files that must each be cleaned and sent back. It also fits teams that cannot put client or employee data into a cloud tool, because the files stay on the PC.
Three kinds of buyer should look elsewhere. Teams on Mac or Linux need a cross-platform tool, since Kramata supports Windows 10 and 11. Workflows that must run unattended on a server, or pull from databases and web APIs, belong on a platform such as KNIME or Alteryx; everything Kramata documents happens inside the desktop app, on files. And analysts who already work comfortably in Power Query inside Excel gain less, because Kramata's main advantage is that it needs no query language.
Strengths & Trade-offs
Pros
- Recipes make a monthly cleanup repeatable and shareable as a .json file.
- Local processing with no account suits sensitive client and employee data.
- Flagged rows keep a reviewable record of what a cleanup set aside.
- Every cleaning rule is in the free edition; batch processing is a $29 one-time purchase.
- Plain-language steps with a live preview suit non-technical users.
Cons
- Supports Windows 10 and 11 (x64) only.
- A young product: first released in June 2026, current version 1.0.3.
- Merge takes the first match per key, so it works as a lookup rather than a full join.
- Pro, the edition with batch processing, is marked Early Access.