Changelog

Added Features

• Physiological Workout Analysis: The AI training assistant now deeply evaluates detected intervals versus planned target structure, cardiac drift (aerobic decoupling), cardiovascular efficiency factor (EF), and pacing smoothness (VI).

Improvements

• Chatbot Response Speed: Optimized the AI reasoning effort to deliver significantly faster and smoother conversational responses.

• Direct Chat Focus: The chatbot now answers questions directly and concisely, eliminating repetitive preambles, canned greetings, and unsolicited congratulations.

Bugs Fixes

• Weekly Summary Emails: Fixed an issue causing duplicate weekly summary emails.

Added Features

• Intervals.icu Sync: Workout modifications, plan replanning, and event adaptations are now automatically synced with your Intervals.icu calendar.

• New AI Engine: Upgraded IA model, delivering training plans with better variety and optimal physiological progression.

Bugs Fixes

• Bike Maintenance Alerts: Fixed component life expectancy alert checks and enabled automatic propagation of maintenance warnings to the parent bike.

Added Features

• Training Assistant (Chatbot): Added persistence for conversation history and chat state, alongside support for granular actions.

Improvements

• Intervals.icu API Client: Developed a rate-limiting client to prevent bottlenecks and rate limit errors when communicating with the API.

Added Features

• Historical Fitness Comparison: Equipped the chatbot assistant with the ability to analyze and inject the last 180 days of training load history (CTL/ATL/TSB) enriched with weekly aggregates, events, race results, and active plan weeks/phases.

• Weekend Foresight: Implemented automatic next-week domain activation on Saturdays and Sundays during general/briefing queries, ensuring the assistant can see upcoming workouts when planning or giving briefings.

Improvements

• Intelligent Recovery Awareness: Added a specific instruction to the system prompt telling the assistant to check the scheduled workouts for today, tomorrow, and the day after, preventing redundant recovery recommendations when rest or light training is already scheduled.

• Coggan TSB Zones: Added the official Coggan TSB zone guidelines (Over-rested, Peak Form, Neutral, Productive Overload, and Critical Fatigue) to the prompt instructions for better performance advice.

Bugs Fixes

• Performance Metrics Consistency: Resolved a discrepancy in CTL/ATL/TSB calculations by ensuring that normalized power (normalizedPower) is correctly selected and passed when calculating the training stress score (TSS) of activities.

Added Features

• Detected Intervals in Activities: Added a new responsive visual card in the activity detail page showing intensity intervals detected automatically by Intervals.icu.

• Interval Badge in List: A compact badge displaying the total interval count is now shown next to each activity in the main list (desktop and mobile layouts).

• Interval Context for Chatbot: The AI training assistant now has access to the detected intervals to better analyze your training intensity.

Improvements

• Power Zones in Chatbot: Fixed a bug in the zone durations formatting that prevented the chatbot from reading power and HR zone distributions correctly.

• Chatbot UI Optimizations: Scrolling improvements, suggestion chips that smoothly collapse when typing, and a collapsible weekly plan view on mobile devices.

Added Features

• Help Guide: The application's documentation has been integrated with the personal assistant, enabling it to provide step-by-step instructions on how to use any platform feature directly inside the chat.

• Conversation Persistence: The chatbot now securely stores message history from the last 7 days and introduces a dedicated button to start a "New Conversation" at any time.

Improvements

• Smart Data Loading: Implemented an on-demand selective data loading architecture that analyzes the athlete's query and retrieves advanced details (such as equipment, goals, zones, nutrition, or help sections) only when relevant, drastically reducing latency and token usage.

• Smart Suggestions: Quick suggestion chips have been expanded and redesigned to compute dynamically based on the athlete's current status and most common queries.

• Message Formatting: Added full support for rendering Markdown headings within chatbot bubbles for a highly readable response layout.

Added Features

• Personal Training Assistant (Beta): Designed and implemented a conversational chatbot on the athlete's Dashboard, allowing natural language interaction to query and manage your calendar.

  • Auto-generated Smart Briefing: Upon first opening, the assistant proactively generates a narrative summary detailing your day's objectives and performance overview in natural language.

  • Native Fitness Metrics Integration (CTL/ATL/TSB): The AI computes your actual 180-day activity history to precisely obtain your Chronic Training Load (fitness), Acute Training Load (fatigue), and Training Stress Balance (form), offering personalized sport-science coaching advice.

  • Dynamic Contextual Suggestions: Replaced static suggestion links with responsive quick chips that adapt dynamically depending on the current day's calendar status (e.g. shortening or moving a workout if pending, starting self-evaluation if completed, or adding recovery sessions on rest days).

Send us your feedback!

Bugs Fixes

• Ad-hoc Workouts without Plan: Resolved a critical application crash when attempting to view or edit the details of a workout not linked to an active training plan (such as standalone ad-hoc calendar entries).

• Meal Planning: Resolved a sorting bug in the nutrition layout on rest weeks where the dinner suggestion was rendered before the afternoon snack.

Added Features

• Event Management: Added a filter to show only active plans in the event modal and implemented read-only restrictions for completed events.

Improvements

• Training Plan Contextualization: The AI context has been optimized to provide a better understanding of how training plans are structured.

Bugs Fixes

• Long Ride Preference: Fixed a bug that prevented the preferred long ride day from propagating from the form to the AI generation logic, successfully ensuring longer workouts on the user's preferred day.

• Volume in Multi-Phase Plans: Fixed a category overlap conflict in the final Taper phase where the taper week duplicated into load weeks, causing the AI to mistakenly reduce weekly hours in advanced phases.

Added Features

• Goal Celebration: Added a brand-new animated and visual celebration experience upon completing an event or goal. When a synced activity matches a scheduled event, an interactive modal is displayed with on-screen confetti animations, a virtual trophy, randomized motivational messages, and a summary of the completed activity.

Bugs Fixes

• AI Modification Menu: Fixed a visual and layout overflow issue with the calendar's popup menu for modifying workouts using AI. The menu is now positioned dynamically and intelligently based on available screen space, and closes correctly when clicking outside.

Added Features

• Completed Events: Events whose dates have passed are now automatically grouped and displayed in a dimmed style within a new collapsible section at the bottom of the events page.

Improvements

• Weekly TSS Chart: The Daily TSS area chart in the advanced analytics section has been replaced with a Weekly TSS bar chart, providing a better overview of the accumulated training load.

• Goal Type: The "Priority" label for events has been renamed to "Goal Type" (Main, Secondary, and Minor Goal) throughout the application for better clarity.

Bugs Fixes

• Event Date Validation: Implemented a restriction that prevents users from creating new events on past dates.

• Training Plan Range Validation: When linking an event to adapt a training plan, the system now strictly validates that the event date falls within the timeframe of the selected plan.