Changelog

v1.4.8 20 July 2026Latest

• A clearer first start: the new compact trust screen explains what leaves your browser, what is never stored, and links directly to the Privacy Policy and Terms.
• Privacy-safe reliability signals: anonymous product events never contain chat text or account data and now survive temporary offline periods without being counted twice.
• Better diagnosis of the real experience: monitoring now distinguishes a completed analysis from an error, usage limit or cancellation after all fallback attempts.
• Internal health probes are separated from real product usage, so reliability decisions are based on user traffic.

v1.4.5 18 June 2026

• Capture fills the panel: grab a conversation with one button — it drops straight into the editor, where you review it, choose the result language and run the analysis.
• A free taste of Pro: on the free plan you get one free run of each Pro style (Technical / Brainstorm / Summary) every week — on top of the regular five, long dialogues included.
• Choose your result language: pick the language of the analysis itself from ~20 options on the capture screen, independent of the interface language. Numbers, names and quotes stay exactly as written.
• "Found in source" badges: key insights that quote your conversation word-for-word are marked as verified against the original text, so you can trust what the brief attributes to the dialogue.
• Sharper analysis styles: the Technical, Brainstorm and Summary presets now produce noticeably more distinct, focused output, with a short hint under each section explaining what it gives you in the current style.
• More faithful to your conversation: stricter rules keep the analysis grounded in what was actually said — standard suggestions are clearly labelled as suggestions, and your own decisions and thresholds are kept verbatim.
• Pro just works: it activates automatically right after purchase and survives reinstalls or a new computer via Chrome's synced storage (nothing extra is sent to us).
• Task-tracker and ADR exports now include each obstacle's priority, and a welcome page with a 3-step quick start opens after installation.

v1.4.3 6 June 2026

• New — "Try a sample": a one-tap button on the start screen runs a built-in example and shows you a real brief in seconds, without opening a chat first.
• Refreshed result screen: color-coded sections (goal, next action, insights, obstacles), the recommended next action now appears on every brief, and a cleaner, easier-to-read solution tree.
• More reliable capture — grabbing a conversation no longer pulls in surrounding page chrome (sidebars, buttons, the composer), so briefs stay focused on the actual dialogue.
• Markdown / plain-text copy and .md download now follow your chosen language instead of always being in Russian.
• The result screen always shows the analysis style (preset) that produced it — including when you reopen a brief from history.
• Steadier extraction of Claude conversations that include interactive widgets.
• Behind the scenes: hardened server-side abuse protection and more resilient analysis.

v1.4.2 29 May 2026

• When no AI chat is open, the side panel now shows one-tap shortcuts to all six supported platforms — jump straight into ChatGPT, Claude, Gemini, Perplexity, Qwen or DeepSeek.

v1.4.1 29 May 2026

• Long dialogues analyze noticeably faster — anything up to ~450K characters now runs in a single pass instead of being split unnecessarily.
• Smarter upgrade screen when you reach the free weekly limit.
• Direct link to the website from inside the extension.
• Cleaner wording in the analysis progress indicator.

v1.4.0 29 May 2026

• Analysis engine upgraded to Gemini 3.1 Pro — a frontier model with a 1M-token context window.
• Resilient multi-provider fallback — you no longer hit "model unavailable" errors.

Telegram bot March 2026

• Buy analysis packs (10 / 50 / 200) right inside Telegram — Telegram Stars or card / SBP in rubles.
• Send bug reports directly to us through the bot.
• File-upload support and more reliable message handling.

v1.3.2 25 March 2026

• More reliable capture of inline AI widgets — no more silent data loss on complex pages.
• Pro licensing via Lemon Squeezy.
• License-input and header layout fixes.

v1.3.1 17 March 2026

• Fixed an issue where long-dialogue analysis could be blocked.

v1.3.0 25 February 2026

• Share any brief by link.

v1.2.0 18 February 2026

• Handle very long dialogues (up to 1M tokens) via map-reduce processing.
• Automatic fallback model when the primary AI is unavailable.

v1.1.0 14 February 2026

• Four analysis presets: Universal, Technical, Brainstorm, Summary.
• Auto-capture from 6 AI platforms: ChatGPT, Claude, Gemini, Perplexity, Qwen, DeepSeek.
• Export to Telegram and Jira formats, plus preset badges in results.

v1.0 10 February 2026

• First release — turn any AI conversation into a one-page decision brief.
• Side panel, analysis history, Russian / English interface.