Eight distributions
Normal, Student's t, gamma, Weibull, log-normal, exponential, Bernoulli and Poisson with PDF, CDF, inverse CDF and moments.
Technology
Built for science, compiled for speed.
A purpose-built desktop platform: a native statistical engine, a unified file layer with cloud sync, and a pipeline from raw data to publication-ready figures. Everything runs on your machine.
70+statistical functions, native
28chart types, one data model
mstypical test latency
0servers between you and a result
Statistical engine
A native engine, numerically careful
Written in Rust and compiled ahead of time. No browser statistics library, no cloud round-trip, and the same answer on every machine.
Descriptive statistics
Mean, median, variance, skewness, kurtosis, IQR, arbitrary quantiles and trimmed statistics, computed with Welford's algorithm and compensated summation.
Parametric tests
z-tests, one-sample, two-sample, Welch and paired t-tests, chi-square goodness of fit and independence, one- and two-way ANOVA.
Non-parametric tests
Mann–Whitney U, Wilcoxon signed-rank, Kruskal–Wallis and Friedman for data that does not meet parametric assumptions.
Assumption checks
Shapiro–Wilk, Anderson–Darling, Kolmogorov–Smirnov and Jarque–Bera normality; Levene, Bartlett and Mood's median for variances, with a recommended test.
Correlation and effect sizes
Pearson, Spearman and Kendall with confidence intervals; Cohen's d, Hedges' g, Glass's delta, Cliff's delta, eta-, omega- and partial eta-squared, Cohen's f.
Power and multiple comparisons
Power and minimum detectable effect for one- and two-sample designs; Benjamini–Hochberg, Benjamini–Yekutieli, Holm and Bonferroni adjustment.
Multivariate and mixtures
PCA, k-means and Gaussian mixture models, with log-normal and multi-peak fitting for distributions that are not one population.
Survival and time series
Kaplan–Meier with confidence bands and Cox proportional hazards; ARIMA, ACF and PACF, Ljung–Box.
Curve fitting
Levenberg–Marquardt non-linear least squares with automatic peak detection and characterisation.
Meta-analysis
Forest plots with fixed and random effects; I², Cochran's Q and τ² heterogeneity.
Outliers and transforms
Grubbs, Dixon's Q, generalised ESD, modified z-score and IQR methods; Box–Cox and Yeo–Johnson with automatic lambda selection.
Workflow
From raw data to publication, in one application
Step 01
Import
Open CSV or Excel, or paste. A multi-tier parser copes with merged cells, formulas, hidden columns and multi-sheet workbooks.
importStep 02
Explore
Column types are detected, descriptive statistics and previews appear before you choose an analysis. No manual casting.
exploreStep 03
Analyze
70+ statistical functions in the native engine. Hypothesis tests, normality checks, correlations, survival: results in milliseconds.
analyzeStep 04
Visualize
28 chart types with real-time editing, from histograms and raincloud plots to survival curves, forest plots and PCA biplots.
visualizeStep 05
Refine
Gaussian fitting, significance brackets and journal templates. Apply a Nature or Cell style and the chart reformats instantly.
refineStep 06
Export
SVG, PDF or PNG up to 1200 DPI, or a .labt file that keeps the full chart state for later editing.
export
File platform
Your research, organised
An Explorer-style file manager built for scientific work, with cloud sync, team sharing and offline access.
Three storage types
Personal cloud storage, team-shared folders and pinned offline files in one interface with tabs and split view.
Offline sync
Pin files or whole folders. Background sync with conflict detection when you reconnect.
Team workspaces
Storage isolated per team with role-based access. Upload, organise and collaborate on data files.
Search and tags
Tag files with experimental metadata and search filenames and tags fuzzily.
Drag and drop
Move between personal, team and offline storage by dragging. Drop files from the desktop to import.
Versions and trash
Autosave with retention, soft delete with restore. Uploads are tracked in the file index the moment they start.
Why native matters
Your data never leaves your machine unless you send it
- Compiled code. The statistical engine is Rust compiled ahead of time, not interpreted JavaScript.
- Parallel. Large datasets are processed across CPU cores.
- Numerically stable. Welford's online moments, Kahan summation, log-sum-exp: the floating-point traps naive implementations fall into are handled.
- No cloud dependency for analysis. Works in airplane mode and behind firewalls.
- Scales from ten rows to ten million without changing tools.
Stack
Tauri 2 and Rust for the desktop shell and engine; React 19 and TypeScript for the interface; Plotly for interactive rendering; Hasura GraphQL over PostgreSQL with row-level security for records; Zitadel for OAuth 2.0 / OpenID Connect; S3-compatible object storage for files; LiveKit for calling; a local JSON-RPC server and MCP adapter for automation.
RustTauri 2React 19PlotlyGraphQLPostgreSQLS3MCP
A different approach. Browser-based tools lean on JavaScript statistics libraries. LabTools runs every computation through a purpose-built native engine compiled into the desktop application, so analyses run faster, handle larger datasets, and give numerically identical results regardless of operating system.
Get started
Run your first analysis in minutes
Download LabTools, drop in a CSV, and see the engine on your own data.