v2.10.0
Language-Aware Link Label Reasoning
The AI now reasons about link labels, ensuring generic administrative descriptors are translated to the response language while preserving official acronyms (like DGSN, CNIE) and brands (like GitHub) in Latin script for Arabic users.
Strict Case Normalization
Automatically normalizes all-caps external service names into Title Case for cleaner readability while keeping real acronyms uppercase.
Guest Limit Prompt Handling
Improved the flow for guest users, displaying friendly prompts when limit thresholds are reached and smoothing the sign-up process.
v2.2.0
Adaptive Retrieval Agent
The search pipeline now behaves like an intelligent agent: it searches, verifies coverage, identifies gaps, adapts its strategy, and searches again — up to 4 rounds — before sending only relevant sources to the answer model. Irrelevant keyword matches are filtered out.
DGSN Police Services Integration
L'Cachet now indexes police administrative services from the DGSN (Direction Générale de la Sûreté Nationale), covering citizen procedures such as theft declarations, police certificates, criminal records, and more.
NARSA Road Safety Services Integration
L'Cachet now indexes road safety and vehicle services from NARSA (National Road Safety Agency), covering speeding sanctions, radar fines, license points, vehicle registration, technical inspections, and more.
Per-Mode Search Round Limits
Search depth now adapts to your mode: Citizen Brief and Standard get 2 rounds, Advanced gets 3, and Lawyer/Research modes get up to 4 rounds of iterative retrieval for deeper coverage.
Smarter Partial Answers
The AI now provides useful partial answers even when no source directly covers your question, citing supporting background sources while clearly stating what is and is not covered by indexed documents.
Copy with Sources
When you copy an AI response, the clipboard now includes all cited legal sources with their titles and URLs, making it easy to save or share complete sourced answers.
Usage Email Shows Percentage
Monthly allowance notification emails now show a clear percentage bar instead of fractional dollar amounts, making your usage easier to understand at a glance.
Faster Parallel Document Retrieval
Vector searches and embeddings now run in parallel across queries, reducing per-round search time by roughly half.
Input Focus & Background Consistency
Clicking the empty space inside the chat input now focuses the text area. The input background matches the suggestion cards for a cleaner look.
v2.0.1
Microphone Access Fix
Resolved an issue where browsers blocked microphone dictation on the production domain due to restrictive permissions policy headers.
v2.0.0
Smart Citation Highlighting & Inspection
Citations in AI responses now highlight on hover with an animated glow. You can inspect, expand, and verify every referenced legal source directly in the chat.
Adjustable Text Size Preferences
You can now increase or decrease text size across the entire interface from your settings panel, making the platform more comfortable for every screen and preference.
Changelog Page in Footer
A dedicated Changelog page is now linked from the footer so you can always stay up to date with the latest features, improvements, and design changes.
Admin User Details & Management
Admins can now view detailed user profiles, inspect account activity, and manage user roles from a dedicated admin dashboard section.
Enhanced Source Links & Citations
Legal sources now include direct external URLs to official procedure texts. Adjacent citations are automatically separated for cleaner reading, and malformed citation markers are handled gracefully.
Smoother Chat Interruption Handling
When a stream is interrupted, the partial response is preserved and clearly marked. Abort signals and snapshot data ensure the UI stays in sync with the backend state.
Refined Sidebar & Landing Page Layout
The sidebar navigation and landing page components have been polished for better spacing, alignment, and visual consistency across all screen sizes.
v1.9.0
Enhanced Search Relevance
Improved the core semantic matching algorithms to fetch highly precise references from legal documents and official procedures, decreasing query noise.
v1.8.0
Voice Dictation & Settings
You can now speak directly to the AI. A microphone option next to the input box allows you to dictate questions in your chosen voice language.
Collapsible AI Reasoning Steps
Added a live visibility collapse toggle for Gemini's deep thinking processes. You can now see the execution logs and response timers.
Spanish & Portuguese Translation Support
Translated major interface sections and prompts to fully support Spanish and Portuguese native users.
v1.7.0
Refunds and Coupon Discount Management
Integrated transaction dashboards for admins to review deposits, process full or partial refunds, manage discount coupons, and apply registration codes.
v1.6.0
Admin User Management console
Developed user list management panels for admins. Superusers can audit users, toggle roles, and inspect customer profiles.
v1.5.0
Edit Messages & Regenerate Responses
Allows editing user prompts to correct mistakes, or telling the AI to regenerate responses with a different legal mode feature.
v1.4.0
Citizen Advanced Mode & Usage Budgets
Introduced advanced assistant modes for handling complex legal questions, bounded by dynamic cost allowances and usage budgets.