Back to home

Changelog

Every update, improvement, and fix to the AdvisorAI platform.

v1.8.0

latest

Team Management API, Usage & API Quotas tracking, Team Settings UI, and enhanced document features.

New

Team Management API - invite members, list members, manage invitations

New

GET /team/members - list all team members with roles

New

POST /team/invites - invite team members via email

New

GET /team/activity - recent team activity feed

New

GET /team/online - currently online team members

New

Usage & API Quotas API - track API consumption

New

GET /usage/stats - current month requests/tokens usage

New

GET /usage/plan - subscription plan details

New

GET /usage/history - recent API call history

New

Team Settings page - UI for managing team members and invitations

Improved

Added Team menu item to sidebar for company accounts

Improved

Postman collection updated with new Team and Usage endpoints

Fixed

Fixed login 500 error by replacing passlib with direct bcrypt

Fixed

Fixed Swagger UI white page by adding jsdelivr.net to CSP

v1.7.0

latest

Enhanced health monitoring, document processing status API, configurable rate limiting via environment variables, and improved Celery worker visibility.

New

/health endpoint now shows Celery worker status and processing queue length

New

GET /documents/processing-status returns counts of uploaded/processing/processed/failed documents

New

Rate limiting now configurable via environment variables (RATE_LIMIT_LOGIN, RATE_LIMIT_ADVISOR, etc.)

Improved

WebSocket endpoint documented in OpenAPI schema

v1.6.2

Fixed Docker build CI/CD - torch CPU package specification issue.

Fixed

Docker build: Changed torch from '2.2.0+cpu' to 'torch' (latest CPU version from PyTorch index)

Fixed

Dockerfile: Added separate torch install from PyTorch CPU index before requirements.txt

v1.6.1

latest

Training label editor, advisor session persistence, improved document status messages, and Pydantic V2 config migration.

New

Training page now has dropdown selector for each sample label (low/medium/high/critical) — fix for 'needs at least 2 distinct labels' error

Improved

Advisor/chatbot now properly loads and persists session history from backend API

Improved

Chatbot/advisor responses distinguish between processing vs processed documents with actionable guidance

Improved

Config.py migrated to Pydantic V2 — DEBUG boolean now parses correctly from environment variables

Fixed

All Python files validated with AST parser — no syntax errors

Fixed

17 pytest tests passing

v1.6.0

latest

Activity analytics API, document reprocess flow, hardened AI prompts, WebSocket retry guard, and RAG answers that acknowledge in-flight documents.

New

GET /analytics/activity returns daily uploads, processed docs, and knowledge entries (7/14/30d) for dashboard charts

New

POST /documents/{id}/reprocess re-triggers stuck uploads using the stored S3 object

Improved

Chatbot/advisor responses now acknowledge uploaded-but-processing docs instead of claiming no data

Improved

Dashboard activity, compliance, and advisor sidebars pull live counts instead of hardcoded placeholders

Improved

WebSocket client has bounded reconnect attempts to reduce noisy console errors when WS is unreachable

Improved

Postman collection includes activity + reprocess endpoints; API descriptions updated

Fixed

Documents upload crash guard when name is missing; upload response maps original_filename correctly

Fixed

AI training enforces multi-class labels client- and server-side to prevent 500s

Security

Chatbot prompt tightened: grounded answers, no fabrication, explicit handling of processing documents

v1.5.0

Admin Panel, forgot password, document delete fix, audit logs, and full API coverage.

New

Admin Panel (/dashboard/admin) — system-wide overview, user CRUD, company list, audit logs, health alerts

New

Super-admin account auto-seeded on startup (admin@admin.com, role: super_admin, no company)

New

Forgot password flow — /forgot-password and /reset-password pages with Redis-backed token

New

Admin API: GET /admin/companies, GET /admin/health-alerts, GET /admin/audit-logs, POST /admin/users, PUT /admin/users/{id}/role, DELETE /admin/users/{id}

New

Audit logs table (migration 006) — tracks all mutations with user, action, resource, IP

New

Training data supports CSV, Excel (.xlsx/.xls), Word (.docx), and TXT uploads

Improved

Document delete button always visible (removed opacity-0 group-hover hiding)

Improved

POST /admin/ml/predict-risk now accepts JSON body {text} instead of query param

Improved

GET /admin/stats accessible to ADMIN role (not just SUPER_ADMIN)

Improved

GET /admin/audit-logs accessible to ADMIN role with graceful fallback if table missing

Improved

Docs page (/docs) tag icons updated to cover all API groups including Admin and Integrations

Improved

Postman collection regenerated from live OpenAPI spec — all 60+ endpoints included

Fixed

Fixed /admin/stats 500 — double .scalar() on exhausted SQLAlchemy result objects

Fixed

Fixed TrainingPage crash — STATUS_CONFIG typed as Record with 'trained' key and idle fallback

Fixed

Fixed mlStatus response normalization — backend {risk_scorer:{}} shape mapped to {status:string}

Security

Admin routes require minimum ADMIN role; SUPER_ADMIN required for destructive operations

v1.4.0

Business Insights, Compliance Calendar, CI/CD pipeline, and security hardening.

New

Business Health Score — aggregate 0–100 score with component breakdown and AI recommendations

New

Compliance Calendar — automatic statutory deadlines (VAT, payroll, GDPR, corporate tax)

New

Document Expiry Tracker — flags contracts and policies nearing expiry

New

API Documentation page at /docs — full endpoint reference with cURL, Python, TypeScript examples

New

GitHub Actions CI/CD — Docker image published to ghcr.io on every push to main

Security

Account lockout after 5 failed login attempts (15-minute Redis-backed lock)

Security

HSTS header enabled in production (max-age=63072000; includeSubDomains; preload)

Security

Content-Security-Policy with per-request nonces, Cross-Origin-Opener-Policy headers

Improved

Dashboard: added Business Health Score and Document Expiry widgets

Improved

Compliance Calendar page with interactive grid, event detail modal, and category filters

Fixed

CI spaCy model download — replaced broken python -m spacy download with direct wheel URL

Fixed

Vercel deployment — standalone output mode is now conditional on NEXT_STANDALONE env var

v1.3.0

stable

Full platform redesign, Help & Support, AI Training, and typed API client.

New

Help & Support page with FAQ accordion, quick actions, system status, and contact form

New

AI Model Training page with live training log, metric cards, and risk predictor

New

Typed API client (lib/api.ts) with graceful fallback when backend unreachable

New

Logo is now clickable — navigates to / on landing page, /dashboard inside app

Improved

Navbar section links use /#anchor when not on the home page

Improved

Nav links (Features, How it works, Use Cases, Pricing) always visible in dark and light mode

Improved

CTA section redesigned — removed orbital rings, added floating stat cards

Improved

Hero headline updated, placeholder text removed

Improved

Documents page: real upload progress bar, per-file status, download and delete actions

Improved

Company page: invite modal with role selector, member management

Improved

Settings page: saves profile via API, dynamic notification toggles

Security

ARIA attributes fixed across all dashboard pages (no hardcoded boolean aria-checked)

Fixed

Favicon updated to custom AdvisorAI SVG logo

v1.2.0

RAG pipeline improvements, rate limiting, and multi-tenant compliance engine.

New

Hybrid search: BM25 + vector similarity with Reciprocal Rank Fusion (RRF)

New

Compliance engine: gap analysis, risk scoring, country-specific rules

New

Sliding-window Redis rate limiting per endpoint and per IP

New

Celery background task queue for document processing (chunk → embed → index)

Improved

SentenceTransformer model cached in Docker image — no runtime download

Improved

pgvector enabled for cosine similarity search on 384-dim embeddings

Security

Refresh token rotation with Redis blacklist

Fixed

Fixed asyncpg connection pool exhaustion under high load

v1.1.0

Chatbot sessions, analytics exports, webhook integrations.

New

Multi-turn chatbot with persistent session history

New

Analytics export in PDF and Excel formats via ReportLab

New

Webhooks: HMAC-SHA256 signed event payloads

New

MinIO integration for local S3-compatible file storage in development

Improved

Dashboard redesign with candlestick compliance chart and sparkline KPIs

Fixed

Document delete now correctly removes all associated knowledge entries

v1.0.0

Initial public release — document upload, RAG advisor, JWT auth, multi-tenant architecture.

New

Document upload and processing pipeline (PDF, DOCX, XLSX, TXT)

New

AI Advisor with cited answers powered by Groq (Llama 3.1 70B)

New

JWT authentication with access + refresh tokens

New

Multi-tenant architecture scoped by company_id

New

PostgreSQL + pgvector for document embeddings

New

FastAPI backend with async SQLAlchemy

New

Next.js 15 frontend with dark/light theme

Want to see what's next? Request a feature or check our API docs.