• Paste or upload a table to get an analysis selected by the pre-specified decision table, with plain-language reasons and a decision path.
• View effect sizes, robustness checks, APA reporting, reproducible R code, and links to the existing statistics tools.
• The analysis stays in the browser; ambiguous table layouts prompt for a small number of design choices.

• Group comparisons now use subjects as the unit of analysis. Welch's t-test and Welch ANOVA are the defaults, with rank tests shown as robustness checks; zero-heavy data use a two-part analysis; technical failures are excluded and listed. Results from earlier versions may differ and are not wrong.

A 2026 study connects mitochondrial protein quality control to oocyte competence during ovarian aging. In mice and a small human oocyte maturation experiment, loss of ClpX increased unfolded proteins and linked to PTGDS/PGD2 signaling, while the experimental inhibitor AT-56 improved several maturation readouts. The findings are mechanistic, not a fertility treatment.

• Section 2 of the Terms of Service now matches what each tier actually includes: every analysis feature is available on every account tier, and the tiers differ only in the daily limit on result downloads (5 for anonymous visitors and Free, 50 for Lab Pro, 500 for Institution)
• The previous table listed dot-blot automatic detection and full quantification as Lab Pro features, and batch analysis and the dot-blot engine API as Institution features; free accounts have had these since June 2026, as announced when Lab Pro opened on 15 July
• Prices and the limit for each tier do not change. The subscription agreement now refers to the Terms for tier features, and seats and any additional Institution API arrangements are set in the order form
• Downloads by a signed-in account now always count against that account's limit: data wash used to count every download against the 5-per-IP anonymous limit, and on all three tools an account fell back to that limit an hour after signing in. The sign-in is now renewed automatically

Several AI CLIs shared a memory system we designed ourselves, in one workspace, for a year. It accumulated about a thousand episodic memories and more than five hundred knowledge notes. What caused the loss was not that the AI forgot. It was that the AI trusted something it had not checked. This piece covers the three fixes in Ghost In Shell 5.2, and six pitfalls we hit ourselves, each with the check we use now.