What's new and what's coming soon

A growth and reliability release that turns referrals into a complete user and admin workflow, adds automatic Advanced Mode journal analysis, and corrects confusing monthly word-count presentation for unlimited plans.

• NewAdded a signed-in referral page with personal referral links, share-card generation, copy actions, reward history, milestones, and badge progress.
• NewAdded Referrals to the logged-in Features navbar dropdown so users can reach the referral command center directly.
• NewAdded an Admin Referrals page with click totals, referral totals, qualification counts, rejection counts, reward grants, reward days, estimated reward value, conversion rates, fraud flags, and recent activity.
• NewUpdated referral incentives to unlock badges at 1, 3, and 5 qualified friends, then Core, Performance, and Platinum plan rewards at 10, 25, 50, and 100 qualified friends.
• NewAdvanced Mode users now receive automatic CBT-inspired thought records, cognitive pattern detection, and deeper personality signal extraction whenever journals are analyzed.
• ImprovedSignup now places referral entry beside account credentials, while onboarding captures what brought each user to Rohy AI immediately before the product tour.
• ImprovedRohy AI now compacts older personality memory into a durable long-term context summary so Personality Twin and Deep Analysis can retain more user history as context grows.
• ImprovedGeneric AI prompts now feed the same personality-learning pipeline as chat, journals, reflections, and clinical responses so user input improves future personalization.
• ImprovedReferral tracking now records clicks, signups, qualified conversions, reward grants, milestone progress, validation audit logs, and privacy-safe fraud signals without exposing journal content.
• FixedMonthly word count now prefers the current-month interaction ledger before falling back to legacy usage counters, preventing stale inflated totals from overriding newer tracking.
• FixedUnlimited-plan usage cards now display "Unlimited plan" instead of a confusing "/ infinity" quota marker.

This phase tightened feature gating, repaired achievement progress, improved onboarding persistence, and removed hydration hazards from the admin profile table.

• ImprovedAdvanced Mode now gates Cognitive Pattern Check, CBT thought records, distortion insights, and Clinical Models consistently across dashboard, account, APIs, and navigation.
• FixedClinical Models only appears in the logged-in Features dropdown when Advanced Mode is enabled, and the page requires Advanced Mode before loading advanced clinical content.
• FixedAdvanced Mode toggling now routes through a server-backed account endpoint so failed profile writes surface clearly instead of logging an opaque empty object.
• ImprovedOnboarding and guided tours now use server-backed state to avoid repeating completed tutorials across devices or browser sessions.
• FixedAchievements now calculate journals, streaks, reports, distortions, thought records, and referral progress from hardened server queries with safe fallbacks.
• FixedResolved the admin profile nested-button hydration error by separating row expansion behavior from per-row export controls.

A feature phase centered on deeper reflection tools and healthier growth loops, with advanced controls for users who want clinical-style models.

• NewIntroduced CBT-style thought records for advanced users, including structured prompts, reframes, and trend-friendly metadata.
• NewAdded cognitive distortion tracking and pattern summaries as an Advanced Mode feature.
• NewAdded social share actions for insight cards, including native sharing and downloadable images suitable for Instagram, Snapchat, TikTok, Facebook, and general sharing.
• ImprovedLanding-page hero sizing now adapts to fill the visible device height beneath the navbar across desktop and mobile viewports.
• SecurityExpanded privacy, disclosure, and safety language to cover AI outputs, provider sharing, referrals, analytics, retention, and non-emergency limitations.

The Rohy AI landing page has been rebuilt into a fuller, more premium conversion journey with stronger product storytelling, restored testimonials, clearer trust messaging, and prominent iOS download paths.

• ImprovedRedesigned the homepage hero around AI journaling, AI reflection, insights, personal growth, and provider collaboration with richer product mockups and clearer CTAs.
• ImprovedAdded a direct trust section covering privacy-focused design, secure data handling, HIPAA-aware workflows where appropriate, and consent-first provider collaboration.
• ImprovedRebuilt the feature journey with alternating product visuals for AI Journaling, AI Chat, AI Insights, Mood Tracking, Personality Twin, Mental Health Overview, Provider Collaboration, Voice Journaling, and the Mobile App.
• NewRestored testimonials with three continuous scrolling lanes and more grounded stories about journaling consistency, self-awareness, pattern recognition, and provider collaboration.
• NewAdded dedicated mobile app, social proof, how-it-works, FAQ, and closing CTA sections with App Store download buttons powered by the shared App Store configuration.

The official Rohy AI iOS app is live, giving users a calmer way to journal, reflect, and revisit insights from iPhone while keeping the web experience close at hand.

• NewRohy AI can now be downloaded directly from the Apple App Store.
• NewMobile journaling is available on iPhone for capturing thoughts when they are still fresh.
• NewAI reflections are now easier to access on the go through the native iOS experience.
• ImprovedWeb and iOS users can sign in with their Rohy AI account for cross-platform access between browser and iPhone.
• ImprovedAdded App Store download CTAs across the website so users can find the mobile app from the homepage, pricing, features, blog, navigation, and footer.

A product-wide polish release that upgrades the visual system, adds a new longitudinal Study workspace for admins, and hardens tutorial persistence and legal disclosures across the platform.

• ImprovedRefined the landing, pricing, and admin experiences with a more premium visual system, stronger hierarchy, smoother motion, and cleaner spacing across major surfaces.
• NewAdded the admin Study workspace with per-user baseline vs. ending mental health scores, engagement metrics, expandable clinical model tracking, and date-range filtering.
• ImprovedPricing cards now present plans in a weekly display format while preserving the underlying monthly and annual billing logic at checkout.
• FixedTutorial completion now respects server-backed account state so completed users no longer see the onboarding flow reappear across devices or browsers.
• FixedCookie consent presentation and several responsive spacing issues were tightened for smaller laptops, tablets, and narrow mobile layouts.
• SecurityPrivacy, terms, and cookie disclosures were updated to align with current analytics, consent, and longitudinal reporting behavior.

A major milestone for clinical integrations. Providers now have a dedicated workspace with exportable progress packets designed to slot into existing session prep workflows.

• NewIntroduced provider-facing progress exports with structured PHQ-9 and GAD-7 trend packets formatted for session prep.
• NewAdded the Clinical Dashboard view with patient caseload overview, risk alerts, and weekly engagement trends.
• NewProviders can now write private clinical notes per patient, saved separately from the patient-visible care plan.
• NewImplemented "Shared View" for Parent/Guardian account types — a curated, safe summary that excludes raw clinical scores.
• ImprovedInvite flow for connecting patients to providers has been redesigned with clearer consent language and per-item sharing toggles.
• ImprovedClinical alert threshold logic now adapts based on the patient's historical baseline rather than fixed cutoff scores.
• FixedResolved a race condition where provider invitations sent within 5 seconds of account creation were occasionally dropped.

We've significantly improved the quality and relevance of AI-generated summaries and completely restructured the Weekly Mind Report format based on user and clinician feedback.

• ImprovedRefined the journal analysis pipeline to produce shorter, more actionable summaries with clearer pattern identification.
• ImprovedWeekly Mind Reports now include a "Key Moments" section that highlights the top 3 emotionally significant journal entries from the week.
• ImprovedAI-detected emotions are now displayed with confidence tiers (High / Moderate / Suggested) to avoid over-stating interpretations.
• NewAdded a "Tone over time" sparkline chart to Weekly Reports that visualises emotional valence across all 7 days.
• FixedFixed an edge case where the life story generator would occasionally duplicate a paragraph when a user had more than 80 entries.
• PerformanceReport generation latency reduced by ~38% through parallel prompt chunking and response streaming.

Voice journaling has been rebuilt from scratch to feel fast and natural. Pattern Rewind lets users scroll back in time and surface recurring emotional themes across any date range.

• NewVoice journaling rebuilt with real-time waveform visualisation, auto-stop on silence, and instant transcription feedback.
• NewPattern Rewind feature: select any 7–90 day range and receive an AI summary of recurring themes, triggers, and growth signals.
• NewAdded "Journal Streak" tracking to the dashboard — shows current streak, longest streak, and a heatmap-style activity calendar.
• ImprovedVoice recordings are now compressed on-device before upload, reducing storage usage by approximately 60% without quality loss.
• ImprovedTranscription accuracy improved noticeably for users who speak quickly or switch between languages mid-sentence.
• FixedFixed a bug on iOS Safari where the voice recording button would become unresponsive after the first session.
• PerformanceDashboard initial load time dropped significantly after lazy-loading the journal heatmap and deferring non-critical API calls.

A targeted patch release focused on eliminating auth refresh issues that affected a subset of returning users and restoring reliable session state after plan upgrades.

• FixedResolved dashboard loading failures that affected some returning users after Supabase auth session refreshes.
• FixedPlan metadata was sometimes stale after a Stripe checkout completion — now immediately refreshed on redirect back.
• FixedFixed "Unauthorized" flash that briefly appeared on the journal page for users with active sessions on slow connections.
• FixedResolved a mobile-specific layout bug on the account page where the subscription section was clipped on small screens.
• SecurityAdded CSRF protections to all state-changing API routes and removed a legacy unauthenticated endpoint no longer in use.
• ImprovedToken refresh logic now silently retries once on failure before surfacing an error to the user, reducing spurious logouts.

Overhauled the pricing presentation for clarity and aligned plan limits to better reflect the real-world workflows of individuals, therapy clients, and clinical providers.

• ImprovedUpdated pricing page with clearer plan comparisons, honest usage limit callouts, and a "Right for me?" quiz widget.
• ImprovedRevised plan limit logic: word counts now track billable words only, excluding system prompts and internal context from quota.
• NewAdded a guided product tour for new signups that walks through journaling, mood tracking, insights, and the persona system in 8 steps.
• NewFree plan users now receive a soft-limit nudge with a breakdown of their remaining quota instead of a hard block error.
• ImprovedOnboarding quiz added to personalise the initial AI model based on whether the user is journaling for self-reflection, therapy, or provider use.
• FixedStripe webhook signature verification was silently failing for some events — now properly surfaced and logged.

The Persona system lets users create distinct AI conversation styles — from a no-nonsense cognitive coach to a gentle reflective companion — each with their own voice and focus area.

• NewLaunched the AI Persona system: create up to 10 custom personas, each with a defined personality, communication style, and focus area.
• NewPersonas now ingest the user's live journal profile for context, so conversations feel genuinely informed by real patterns and history.
• NewAdded a "Persona Gallery" with 6 pre-built starter personas including a CBT coach, a stoic mentor, and a gentle reflective guide.
• ImprovedChat interface redesigned with message reactions, copy-to-journal functionality, and collapsible long AI responses.
• ImprovedPersona context window now smartly prioritises the most emotionally salient journal entries rather than always pulling the most recent.
• NewAdded "Conversation summaries" — AI-generated recaps of each chat session saved to the user's journal timeline for future reference.
• FixedFixed a bug where persona names containing apostrophes broke the chat interface title rendering.

Corrective release focused on accurate PHQ-9/GAD-7 score computation and fixing display inconsistencies in the wellbeing trend charts.

• FixedPHQ-9 severity bands were incorrectly labelling scores of 10–14 as "Severe" instead of "Moderate" — corrected across all displays and exports.
• FixedGAD-7 trend chart was using UTC boundaries for "today" rather than the user's local timezone, causing the chart to update at unexpected times.
• FixedHistorical snapshots were occasionally omitting the final entry of a month when timezone-aware monthly grouping was applied.
• ImprovedMental health score badges now include a plain-language label (e.g. "Mild anxiety") alongside the numeric score for accessibility.
• PerformanceSnapshot computation has been moved server-side and cached per-day, reducing dashboard query load by over 50%.

Two major features rooted in longitudinal insight: a continuously-updated personality map that learns from your writing, and a first-person life story synthesized from your journal history.

• NewLaunched the Personality Map: a live profile of traits, emotional tendencies, cognitive patterns, and interests derived from journal entries over time.
• NewLife Story Generator: produces a first-person autobiographical narrative from a user's full journal history, updated on demand.
• NewPersonality Twin feature: see how your trait profile compares to archetype profiles across openness, resilience, and self-reflection depth.
• ImprovedJournal analysis pipeline upgraded to extract not just emotions but also recurring cognitive distortions, coping patterns, and stated goals.
• NewAdded "Insight Cards" to the dashboard: bite-sized AI observations about the user's current emotional state, updated after each new journal entry.
• ImprovedPrivacy control added to the personality map — users can redact specific entries from contributing to their profile at any time.
• FixedLife story word count is now correctly tracked against the monthly quota, resolving an unintentional free-tier bypass.

The first major iteration of Rohy AI's clinical sharing model — built around explicit consent, granular controls, and the clear principle that raw journals are never shared by default.

• NewLaunched secure sharing: users can invite clinicians or guardians and control exactly what they can see — summaries, scale exports, or selected journal insights.
• NewSharing recipients receive a read-only portal view with no ability to message, modify, or export beyond what the user explicitly enables.
• NewAuto-share mode: optionally deliver the latest Weekly Mind Report to all accepted connections as soon as it is generated.
• ImprovedSharing consent flow redesigned with per-item checkboxes instead of a single "share everything" toggle, based on early clinical feedback.
• NewAdded sharing audit log: users can see exactly when each connected person last accessed their shared data.
• SecurityAll shared portal views are served via scoped, time-limited access tokens that expire after 30 days of inactivity.
• FixedRevoke access was not immediately invalidating the recipient's session token — now terminates access within seconds of revocation.

A full product and technical rebuild. Rohy AI 2.0 moves to a modern App Router architecture, introduces the new design system, and lays the infrastructure for all clinical and AI features to come.

• NewRebuilt entirely on Next.js 16 App Router with TypeScript strict mode, replacing the previous Pages Router codebase.
• NewNew design system with a cohesive dark mode, custom CSS variables, and a glassmorphism card system used consistently across all surfaces.
• NewSupabase Auth replaced legacy custom JWT system — faster session management, built-in OAuth, and a path to MFA.
• NewMigrated all user AI profiles to a vector-indexed Supabase structure to support semantic retrieval in future chat and insight features.
• ImprovedDashboard performance overhauled: server-side data fetching, streaming UI, and skeleton loading states replace the previous client-side waterfall.
• NewIntroduced the journal editor with rich-text support, word count tracking, and real-time AI analysis on save.
• NewStripe integration added with webhook-backed plan sync, ensuring subscription state stays accurate even when browser tabs are closed during checkout.
• PerformanceCore Web Vitals improved significantly: LCP down by ~55%, CLS reduced to near-zero with layout reservations, and INP optimised through deferred hydration.

• FixedJournal entry autosave was triggering on every keystroke — now debounced at 2 seconds to reduce API calls and prevent race conditions.
• FixedMobile Safari keyboard layout shift when the emoji picker opened caused the text cursor to jump — resolved by locking scroll position before open.
• FixedMood check-in widget was not rendering correctly in landscape mode on iPhone SE and similar small-viewport devices.
• ImprovedAdded haptic feedback on mobile for key interactions: completing a journal entry, saving a mood check-in, and receiving an AI insight.
• PerformanceReduced JavaScript bundle size by 18% through tree-shaking unused icon imports and moving chart components to dynamic imports.

Introduced a lightweight daily mood check-in system that doesn't require a full journal entry — making consistent emotional tracking accessible even on the hardest days.

• NewDaily mood check-in widget: rate overall mood, energy, and anxiety on simple 1–10 scales with an optional one-liner note.
• NewMood history chart on the dashboard showing a 30-day rolling view of emotional patterns with peak and trough annotations.
• NewCheck-in streak system: consecutive daily check-ins are tracked and surfaced as a motivational metric on the dashboard.
• ImprovedAI journal analysis now incorporates same-day mood check-in data to produce more context-aware summaries.
• NewAdded push notification support for daily check-in reminders — time and frequency configurable from account settings.
• FixedResolved a timezone bug where users in UTC+ zones were seeing yesterday's check-in prompt instead of today's.