Shipping, visibly.

2026-08-24

• +Seven engines: DeepSeek and Mistral now answer through their own APIs instead of a reseller, faster and on the real production models
• +Raw Llama dropped: its real consumer surface (Meta AI) answers differently, so measuring the bare API model measured a surface no buyer actually uses
• +Scan reliability: extraction runs in parallel and tolerates non-Latin replies, so Cyrillic, Arabic and CJK scans finalize cleanly instead of stalling

2026-08-22

• +Ask Saymetry: natural-language answers over your own scan data, in the dashboard
• +Page-level citations: the exact URLs the engines read, not just domains
• +Working queue: Save / Done / Dismiss on every action-plan item; the next scan measures what you changed
• +This Week's Movers: biggest gains and drops between weekly runs
• +Password sign-in with server-side lockout; magic links stay as fallback
• +Product tour with real screens replaces the public sample report
• +Trust & security page; public changelog (this page)

2026-08-21

• +8 engines on every plan: Gemini (Google-Search-grounded), DeepSeek, Mistral and Llama join ChatGPT, Claude, Perplexity and Grok
• +Sampling honesty: a stability check re-runs part of every scan; reports show mention probability with a 95% interval and pass-to-pass agreement
• +Weekly tracking with alert emails; Slack and Telegram delivery included
• +Competitor battle cards, open-ground map, publish-here target list, and market opportunities in every report
• +Sentiment per answer and a rank headline (#N of M names)
• +Answer-tone, rank and per-engine breakdowns in the dashboard; CSV export
• +Agency partnership (white-label, per-client) and the Content Engine add-on
• +Wizard checkout; free AI-Ready check with emailed full reports

2026-08-20

• +Saymetry launches: the $99 8-engine AI Visibility Scan with website audit, on saymetry.com