Glossary
Key terms and definitions used across the Qubitz AI platform.
| Term | Definition |
|---|---|
| Agent Activity Tree | Real-time panel in the Test Bed showing the live orchestration tree during a conversation - each agent's status, tool calls, reasoning steps, and responses, colour-coded per agent |
| Agentic System | Control Hub tab where the Workflow Architect generates and manages the multi-agent architecture for your use case. Produces 3-7 specialized agents each with a role, model, tools, memory, and system prompt. Deployment to AgentCore fires automatically once the spec is finalized |
| Agentic Workflows | Control Hub tab where the Workflow Architect designs multi-step workflow plans for your use case. Each workflow has a name, description, and ordered step list. Cap: 30 per project, 300 KB total, 24 KB per workflow |
| AgentCore | AWS Bedrock AgentCore - the managed runtime that hosts deployed Qubitz agent systems. Handles orchestration, memory, tool execution, and WebSocket connections |
| AI Assessment | The AI readiness route in the Use Case Tool, accessed via the AI Assessment nav link. After clicking Initiate assessment, choosing Yes launches an adaptive Q&A of approximately 8-16 questions across 6 categories: Organization & Strategy, Tech Infrastructure, Data Readiness, Team & Skills, Compliance, and AI Initiatives. On completion, 6 scored PDF reports are generated: Revised Priority Use Cases, Platform Decision, Gap Mitigation Plan, AWS Fit Feasibility Report, ROI Modeling Report, and Well-Architected Review |
| AI Screening | A Horizontal blueprint that conducts automated voice-based candidate interviews using Amazon Nova Sonic. Evaluates candidates across configurable criteria and delivers structured assessments with scores |
| Alignment Score | AI-calculated score (0-1) measuring how well a use case aligns with the company's stated strategic intent |
| App Builder | The conversational interface shown after login. Describe the business problem you want to solve, and Qubitz guides you through shaping your idea - project name, workflows to automate, target users, data sources, integrations, and success criteria |
| APT (Agentic Process Transformation) | Post-assessment view accessible from the Continue to APT button on the Jamming Board. Maps business processes against the generated use cases and identifies workflows that can be automated with agentic AI. Starts generating in the background when the full AI Readiness route is chosen |
| App Deployer Agent | Platform agent that handles the end-to-end deployment pipeline - provisions infrastructure, builds and pushes containers to AgentCore, and reports deployment status |
| App Refiner | AI-assisted feature development tool inside Application Management that lets you describe changes to your deployed application in plain English and deploy them at the same URL |
| Application Management | Post-deployment dashboard with 10 tabs: Overview, Agents, App Config, Test Bed, App Refiner, Identity & Access, Observability, Compliance & Security, FinOps, and Settings |
| Artifacts | Generated documents stored in the project's Artifacts panel - includes AI Discovery Report, Deep Research Report, Use Case List, Executive Presentation, and AWS Architecture Diagram |
| Best Fit | Green badge auto-assigned to the use case with the highest combined rating and sentiment score on the Jamming Board |
| Blueprint | A pre-built, production-ready multi-agent agent system for a high-value enterprise use case. Qubitz provides 17 blueprints across 5 categories: Horizontal, Healthcare & Life Sciences, Financial Services & Insurance, Media & Entertainment, and Manufacturing & Automotive. Each blueprint is customized to your company via a setup wizard and deployed directly to AWS |
| Brand Kit | Visual identity assets (application name, tagline, primary color, font style, logo) configured during app generation to apply your company's look and feel to the generated frontend application |
| Business Case | Section in the Control Hub Deploy tab providing cost estimates across Conservative, Base, and Growth scenarios, an AWS Architecture Diagram, and PDF export |
| Business Email Validation | Platform-level check that blocks personal email domains (Gmail, Yahoo, Hotmail, Outlook, iCloud, and others) at sign-up, collaborator invite, and the Use Case Tool form. A business or organisation email is required |
| Channel Integration | Configuration in App Config that connects your deployed agent to communication channels - Web Widget, Slack, WhatsApp, Microsoft Teams, and Voice / Calling (via Twilio + Nova Sonic) |
| Connection Log | Real-time scrollable log in the analysis modal showing each agent's actions as the multi-agent pipeline executes |
| Control Hub | The five-tab workspace for designing, testing, and deploying multi-agent systems. Tabs: Agentic Workflows, Agentic System, Data, Test Bed, Deploy. Accessed via the Design App button or from the Blueprints page |
| Customer Support | A Horizontal blueprint that deploys an omnichannel AI support agent across web chat, Slack, WhatsApp, and voice. Handles customer intents, takes actions in connected systems, and escalates to a human when needed |
| Data Tab | Control Hub tab with sub-sections for feeding agents with domain knowledge: Knowledge Base (document upload), Structured Dataset (tabular data), Connections (MCP integrations), and GraphRAG (domain knowledge graph) |
| Deep Analysis | The multi-agent research pipeline that runs after Initiate assessment. Takes approximately 8 minutes (use cases only) or 12 minutes (full AI Readiness route). Produces the AI Discovery Report, Deep Research Report, and Use Case List. The full AI Readiness route additionally generates 6 scored PDF reports |
| Deploy Tab | Control Hub tab that unlocks after first agent deployment. Contains two inner tabs - Generate App (full-stack frontend code generation + Amplify deployment) and API (endpoint URL, key management, Quick Start code, Developer Guide) - plus the Business Case section |
| Design App | Button on the Jamming Board that creates an Application record and opens Control Hub for a selected use case |
| Design Doc Agent | Platform agent that generates a comprehensive Application Design Document for deployed applications covering architecture, data models, API contracts, and implementation details |
| DevGuide Agent | Platform agent that auto-generates versioned developer documentation for deployed projects |
| DPO Pack | A set of 7 GDPR compliance documents generated automatically under the Compliance & Security tab: Privacy Policy, Data Flow Map, Security Policy, Retention Policy, Subprocessor Register, Cookie Policy, and GDPR Controls |
| Evaluation | Structured testing capability in the Test Bed. Run up to 8 metrics across Response Quality (Correctness, Helpfulness, Coherence, Faithfulness, Relevance, etc.), Task Completion, Tool Usage, and Safety dimensions |
| Feature Engineering Agent | Platform agent in Control Hub that handles feature extraction and data ingestion planning for the generated architecture |
| FinOps | Financial operations tab in Application Management - tracks cost and usage of deployed applications by AWS service with daily cost trends and period comparisons |
| Generate App | Action in the Control Hub Deploy tab that streams full-stack frontend code in real-time, auto-pushes to GitHub, and triggers an Amplify build. On success a Live Deployment card appears with the live application URL |
| GraphRAG | Domain knowledge graph builder in the Control Hub Data tab. Extracts entities and relationships from uploaded documents, builds a property graph visualized on Cytoscape.js, and exposes the graph as an agent tool for structured relationship queries at runtime |
| Hyperpersonalisation | A customer intelligence capability that analyses data across 7 dimensions (WHO, WHAT, WHEN, WHERE, HOW MUCH, WHY, RISK) and produces personalized campaign strategies and reports |
| IDP (Intelligent Document Processor) | A Horizontal blueprint that runs an end-to-end document pipeline - classifying, extracting structured data from, validating, reviewing, and reporting on uploaded documents automatically |
| Impact Badge | Color-coded classification (High Impact = red, Medium Impact = yellow) assigned to each generated use case on the Jamming Board |
| Jamming Board | The collaborative workspace for a project. Displays generated use cases as cards with ratings, impact badges, and alignment scores. Team members rate, comment, and rank use cases. The top-ranked use case gets a Best Fit badge |
| Knowledge Base | Document store within the Control Hub Data tab. Upload PDFs, DOCX, and other files that agents can query via RAG (Retrieval-Augmented Generation) at runtime |
| Lending Case Processor | A Financial Services blueprint for end-to-end UK mortgage and loan case processing. 7 agents handle classification, extraction, validation, affordability assessment (MCOB 11.6), chase, and exception routing. Outputs a Decision-Ready Pack PDF with full FCA compliance |
| Live Deployment | Card that appears in the Control Hub Deploy tab after a successful app build. Shows the live application URL and marks the application as deployed |
| MCP (Model Context Protocol) | Open protocol for connecting AI agents to external tools, data sources, and APIs. Configured in the Control Hub Data tab under Connections and managed via the AWS Bedrock AgentCore Gateway |
| Nova Sonic | Amazon's real-time voice AI model used to power voice-based agents including phone interviews, voice support channels, and in-browser voice interactions |
| Observability | Tab in Application Management showing Sessions, Traces, Latency, Error Rate, Token Usage per model, Resource Consumption, and Agent Spans |
| Pipeline | The multi-agent orchestration system that processes business context through sequential phases during Use Case Tool analysis |
| PPTX Agent | Platform agent that generates the Executive Presentation artifact - a PowerPoint deck summarizing use case findings for stakeholders |
| Priority Score | Calculated score in the AI Discovery Report combining Business Value, Feasibility, Risk, and other dimensions |
| Project ID | Auto-generated identifier for each saved analysis project (format: company-slug-numeric-id) |
| RAG (Retrieval-Augmented Generation) | Technique where agents retrieve relevant content from a connected knowledge base at runtime before generating a response, grounding answers in your own documents and data |
| RunChannel Buffer | Server-side replay buffer for UCT WebSocket sessions. Holds up to 400 frames; replays up to 60 frames on reconnect so in-progress analysis resumes transparently after a connection drop. Reconnect window: 25 minutes |
| Setup Wizard | The multi-step configuration flow shown when selecting a Blueprint. Collects company context, domain-specific requirements, and integration targets. Answers are used by the AI to customize the blueprint before deployment |
| Socket.IO | Real-time communication layer used across the Qubitz platform for streaming agent responses, deploy logs, and live status updates between the backend and the browser |
| Strands Agents | The open-source Python multi-agent orchestration framework powering the Qubitz AI backend. Handles agent-to-agent delegation, tool execution, and streaming outputs |
| Test Bed | Live chat environment within Control Hub and Application Management for testing deployed agents - includes Agent Activity Tree, structured Evaluation, and voice mode |
| UCT (Use Case Tool) | The main discovery engine. Accessed via the AI Assessment nav link, it uses a chat-based intake with a Business Requirements panel to collect company context. Results land on the Jamming Board |
| WAFR | Well-Architected Framework Review - automated AWS compliance report generated against a deployed application's infrastructure, covering operational excellence, security, reliability, performance, and cost optimization |
| Workflow Architect | The AI agent powering both the Agentic Workflows tab (workflow design) and the Agentic System tab (agent design) in Control Hub. Runs via WebSocket with two scopes: "workflows" and "agents" |
| Workflow Cap | Per-project storage limit for the Agentic Workflows tab: 30 workflows maximum, 300 KB total across all workflows, 24 KB per individual workflow. The platform warns before a save would exceed the cap |