Release Notes

Version history and changelog for the Qubitz AI platform.

September 2026

Use Case Tool - Chat-Based Intake

  • Intake for AI use case discovery is now fully chat-based with a Business Requirements left panel, replacing the previous form (company name, website, description fields)
  • Business Requirements panel tracks five groups: SIGNED IN, COMPANY, COMPANY PROFILE, YOUR CHALLENGES (HIGH-SIGNAL), and OPTIONAL
  • Platform auto-researches the company on description and pre-fills the panel through natural conversation
  • Initiate assessment button launches a route selection dialog instead of triggering Deep Analysis directly

Use Case Tool - Assessment Route Selection

  • Clicking Initiate assessment now shows a route selection dialog: Yes (full AI Readiness, ~12 min) or No (use cases only, ~8 min)
  • Yes path: runs adaptive Q&A of 8-16 questions across 6 categories, generates use cases grounded in answers, plus 6 scored PDF reports (Revised Priority Use Cases, Platform Decision, Gap Mitigation Plan, AWS Fit Feasibility, ROI Modeling, Well-Architected Review)
  • No path: generates use cases directly - no questions, no extra reports
  • Both routes end on the Jamming Board

Jamming Board - New Actions

  • Continue to APT button added to the Jamming Board header - launches Agentic Process Transformation view
  • Logs button added to the Jamming Board header - view the pipeline execution log for the current run
  • Globe icon on each use case row - links to web research and external references gathered for that use case
  • APT starts generating in the background when the full AI Readiness route is chosen, so it is ready when you navigate to it

August 2026

Identity & Access

  • Identity tab in Application Management is now live - SSO connections, SCIM provisioning, directory sync, domain verification, sign-in policies, grants, and people management
  • Agents in deployed applications can authenticate as trusted service principals with scoped grants
  • Add a GitHub collaborator to a deployed app's repository directly from Application Management without redeploying

Control Hub - Android Mobile App

  • AppDeployer now generates an Android mobile app alongside the web app via Capacitor

GraphRAG - Domain Property Graph Builder

  • New GraphRAG feature inside Control Hub: build, visualize, and evolve a domain knowledge graph for your use case before deploying agents
  • Interactive graph canvas powered by Cytoscape.js - nodes, edges, and evidence clusters rendered in real-time
  • Domain ontology with axiom enforcement - define entity types and relationship rules before extraction runs
  • Predefined safety stages aligned with the KG-validation and guardrails framework
  • Mermaid diagram export alongside the interactive view
  • Accessible via the Data tab in Control Hub

BrandKit - Figma Integration

  • Connect your Figma account to pull brand assets directly into the Brand Kit

Control Hub - Workflow Cap

  • Workflows per project capped at 30, total storage per project at 300 KB, and each individual workflow at 24 KB
  • The UI reports remaining capacity before you reach the limit
  • If a project is full, the platform says so clearly before the save fails

Control Hub - Architect UX

  • Workflow architect now covers the full Agentic System tab - both workflow editing and agent configuration are handled in the same panel
  • Removed separate context dialog - context awareness is inline

Use Case Tool - WebSocket Resilience

  • WebSocket reconnect via attach action - if your connection drops mid-analysis, reconnecting replays the last 60 frames of the pipeline log
  • 400-frame in-memory buffer (RunChannel) keeps the full session alive for up to 25 minutes after completion
  • Pipeline sessions persist for 15 minutes post-completion so a page refresh doesn't lose your results

Use Case Tool - UX Improvements

  • Assessment answers are preserved in session storage if you cancel or navigate away
  • Re-submitting with a prior run in progress gives you a clear choice: Continue the existing run or Start Fresh

Authentication - Business Email Required

  • Personal email domains (Gmail, Yahoo, Hotmail, Outlook, iCloud, and 11 others) are now blocked at signup, team invites, collaborator dialogs, and the Use Case Tool invite form
  • Error is specific: "Please use a business email address. Personal email domains are not allowed."

Version 3

Control Hub - Restructured Tabs

  • Control Hub now has 5 tabs: Agentic Workflows, Agentic System, Data, Test Bed, and Deploy (previously 3 tabs)
  • Agentic Workflows tab - New dedicated tab for designing multi-step agentic workflow plans. Each workflow has a name, description, and ordered step list
  • Data tab - New dedicated tab with four sub-sections: Documents, Structured Data, Connections, and Graph (GraphRAG)
  • Deployment now fires automatically once the Workflow Architect finishes generating the agent spec - the manual Approve button is removed

Control Hub - Workflow Architect

  • The Workflow Architect agent now powers both the Agentic Workflows tab (scope: workflows) and the Agentic System tab (scope: agents) via a unified WebSocket handler
  • Workflow cap enforced per project: 30 workflows, 300 KB total, 24 KB per workflow
  • Platform warns before the cap is hit - users can delete unused workflows before saving

Control Hub - GraphRAG

  • New Graph sub-section in the Data tab for building domain property knowledge graphs
  • AI extracts entities (nodes) and relationships (edges) from uploaded documents
  • Interactive Cytoscape.js canvas for graph visualization with color-coded node types and labeled edges
  • Editable Mermaid pipeline for adjusting the extraction process
  • Built graph is available as an agent tool in the Agentic System tab for structured relationship queries

Use Case Tool - Resilience

  • WebSocket reconnect/replay - If the connection drops during Deep Analysis, the platform automatically reconnects and replays up to 60 of the last 400 frames from the RunChannel buffer. In-progress analysis resumes transparently without data loss
  • Assessment preservation - Partial results and assessment state are preserved across reconnects
  • Upload step continue/fresh - On returning to an analysis with existing uploads, users can choose to continue from the previous run or start fresh

Platform - Business Email Validation

  • Personal email domains (Gmail, Yahoo, Hotmail, Outlook, and 10+ others) are now blocked at signup and when inviting collaborators
  • Business or organisation email addresses are required across all entry points: sign-up, invitation dialogs, and invite forms

Version 2

Blueprints

  • Blueprints page now publicly accessible -- browse and deploy pre-built agent systems without starting from scratch
  • Customer Support Blueprint -- omnichannel support agent with web chat, Slack, WhatsApp, and voice (Twilio). 7-step setup wizard captures channels, intents, knowledge sources, integrations, escalation rules, tone, and languages. AI customizes the full agent system to your company before deploying
  • AI Voice Screening Blueprint -- automated voice candidate interviewing via Amazon Nova Sonic. Configurable role types, interview stages, evaluation criteria, and ATS integrations
  • Hyperpersonalisation (HP) Blueprint -- customer intelligence and campaign personalization pipeline deployable as a standalone blueprint with ECS Fargate ML execution
  • Intelligent Document Processing (IDP) Blueprint -- end-to-end document pipeline (classification, extraction, validation, human review, reporting). A2I onboarding wizard collects document types, extraction schemas, and confidence thresholds

Voice / Nova Sonic

  • Voice-anywhere: Amazon Nova Sonic voice agent can now be bundled with any application, not just dedicated voice blueprints
  • Test Bed voice toggle -- test voice conversations directly in the browser with live transcripts
  • Twilio integration -- inbound/outbound phone call routing with warm transfer to human agents
  • Voice channel configuration per agent in the Agents tab

Control Hub

  • Brand Kit -- upload your logo, color palette, and brand guidelines to apply your visual identity to generated applications
  • Terms of Service acceptance step added to auth and App Refiner flows
  • Chat updates (chat_updates action) -- modify your agent spec via natural language after initial generation, changes streamed in real-time

Application Management

  • New tabs added: Compliance & Security, Identity & Access, Business Case
  • Webhooks configuration -- set up inbound and outbound webhook endpoints for your application
  • DPO Pack generation -- 7 compliance documents (Privacy Policy, Data Flow Map, Security Policy, Retention Policy, Subprocessor Register, Cookie Policy, GDPR Controls)

Infrastructure

  • HP ECS containers hardened: read-only root filesystem, Security Hub controls compliance
  • HP ML pipeline remastered: 8 vectorized modules, 24 dead modules removed
  • Auto-injection of HP environment variables into Amplify apps on deploy

IDP (Intelligent Document Processing)

  • Async document processing with AWS Textract
  • Bedrock Data Automation (BDA) integration for structured extraction
  • Guardrails for content safety on processed documents
  • Human-in-the-loop review (A2I) for low-confidence results
  • Webhook support -- inbound triggers and outbound result delivery
  • Metrics dashboard for throughput, error rates, and cost-per-document tracking

Security

  • HP infrastructure hardened against AWS Security Hub controls
  • Payload-sourced infra config rejected in ECS trigger to prevent injection
  • Private reports bucket with presigned URLs and job ownership validation

Version 1.1

New Module: AI Assessment

  • Standalone AI Readiness Assessment with 3 input modes (Use Case Project, Standalone, New Standalone)
  • Generates 6 scored reports: Revised Use Cases, Platform Decision, Gap Mitigation, AWS Fit Feasibility, ROI Modeling, and Well-Architected Review
  • WAFR (Well-Architected Framework Review) -- 4-step guided review across 6 AWS pillars with dynamic questions, AI auto-fill of 57 WAFR questions, and compliance report generation
  • Accessible from the Dashboard alongside Use Case Tool and Control Hub

Use Case Tool

  • AI Assessment added as standalone module with AI Readiness Assessment (6 categories, 6 scored reports) and previous assessment history
  • Use Case Cards redesigned with vertical timeline layout, expandable card details (Business Value, Implementation Timeline, Estimated ROI, Impact Score, Sentiment Score)
  • Custom Use Cases updated with Business Value, Implementation Timeline, and Estimated ROI fields
  • Artifact Viewer redesigned -- left panel for documents, right panel for artifact list with version dropdowns, Chat button for AI chat mode

Control Hub

  • Agentic System -- 24 models now available across Anthropic, Amazon Nova, Meta Llama, Mistral, NVIDIA, Qwen, MiniMax, OpenAI OSS, Z.AI. New tools: Generate Report, Text to Speech, Speech to Text, Video tools. Agent-as-a-Tool configuration. Platform Agents toggle. Bedrock API key validation on approve.
  • Test Bed -- Live chat with Agent Activity Tree showing real-time sub-agent orchestration. Evaluation system with up to 8 metrics across Response Quality, Task Completion, Tool Usage, and Safety dimensions.
  • Deploy -- Generate App with real-time code streaming and auto-build via GitHub + Amplify. App Refiner for AI-assisted changes with Deploy Changes button. API tab with Developer Guide generation. Business Case dialog with Cost Estimates (3 scenarios), AWS Architecture Diagram, and Export as PDF with AWS Pricing Calculator link.
  • Application Management -- Post-deployment dashboard with tabs for Overview, Agents, App, Business Case, Compliance & Security, Cost & Usage, FinOps, Identity & Access, Observability, and Settings.

New APIs

  • /devguide -- Developer guide generation (WebSocket, env-var keys only)
  • /design-doc -- Design Document generation via presigned WebSocket URL
  • /wafr -- Well-Architected Framework Review (synchronous HTTP)

New Agent Runtimes

  • Design Doc Agent -- 13-section HLD document with 4 appendices
  • SOW Agent -- 11-section Statement of Work with e-signature block
  • WAFR AI Agent -- Automated AWS Well-Architected Framework Review
  • DevGuide Agent -- Auto-generated developer documentation
  • App Deployer Agent -- Full-stack deployment orchestration
  • Business Case Agent -- Infrastructure cost estimation and ROI analysis

Platform Updates

  • Navigation & Layout -- Added Pricing, Contact Us, and Projects sections
  • Technology Stack -- Complete rewrite with Strands Agents SDK, ECS Fargate ARM64, and 24 agent runtimes
  • Glossary -- Expanded from 13 to 25+ terms
  • Release Notes page added

Version 1.0

Use Case Tool

  • Multi-agent research pipeline with Deep Analysis (6-8 minutes)
  • AI Discovery Report (16 pages), Deep Research Report (25+ pages), Use Case List (6 pages)
  • Jamming Board with star ratings, threaded comments, sentiment analysis, Best Fit badge
  • Executive Presentation (10-slide PowerPoint) and AWS Architecture Diagram generation
  • Artifact versioning and regeneration with custom instructions
  • AI Chat Assistant for contextual document Q&A
  • Custom use cases, invite up to 10 collaborators per project
  • Multi-Agentic Pipeline and Execution Phases documentation

Control Hub

  • Agentic System with visual agent builder and natural language refinement
  • Model selection: Claude Opus 4.5, Claude Sonnet 4.5, Claude Haiku 4.5
  • Tool integrations: RAG, MCP, Web Search with 16 pre-configured providers
  • Memory configuration: Short-term, Long-term, Semantic
  • Feature generation and application preview
  • One-click deployment to AWS infrastructure
  • API endpoint generation with key management

Platform

  • Introduction with Quick Start Guide, Authentication, Navigation & Layout
  • Qubitz Workflow, Platform Overview, Core Concepts
  • Technology Stack, Qubitz APIs (7 WebSocket endpoints), Glossary