Changelog — what's new

Every feature, security improvement, and bug fix we ship — dated, versioned, and explained.

October 25, 2026 · v2.8.0 · Sources, numbers and inputs you can check

  • Benchmark citations now say "Quantis modeled benchmark".
  • The result header no longer shows a fixed run time or a fixed AI-agreement score; both appear only when measured.
  • Location coverage is stated as it is: 15 US metros plus a US-average baseline.
  • Hiring decisions typed on the homepage now open the hiring model, which compares cash with and without the hire, instead of being analysed as a new business.

October 24, 2026 · v2.7.0 · Monte Carlo on every report, saved free runs, sample Decision Memo

  • Every report now shows a monthly profit range (low, typical, high) from 1,000 simulated months, with its assumptions printed next to it.
  • A sample Decision Memo shows exactly what the Founder Decision Review delivers, worked through on a fictional business.

August 27, 2026 · v2.6.0 · Benchmark reconciliation + export safeguards

  • Exporting a report that failed an internal consistency check now requires an explicit confirmation before download, instead of silently allowing a normal-looking export.
  • Published a formal security-disclosure policy (security.txt) so security researchers have a documented, direct way to report issues.

Methodology · Status · Security