Skip to content

ILANIT Tarot – Functional & Technical Solution Proposal

Executive Summary

The “ILANIT | Precise Answer” initiative is a digital emotional-support and self-awareness experience, delivered via WhatsApp.
It merges human warmth, positive psychology, and AI assistance, providing users with instant, empathetic responses that feel personal yet responsible.
The goal is to make emotional guidance accessible, non-mystical, and authentic, reflecting Ilanit’s unique communication style.

Users interact with a WhatsApp menu featuring key life areas (Love, Health, Career, Finance, Emotions) and receive curated or AI-generated replies inspired by Ilanit’s tone and philosophy.
The system encourages continued engagement through soft calls-to-action (CTAs) such as “Ask a personal question” or “Subscribe to daily guidance messages.”

Revenue is generated through:

  • One-time paid questions (₪19 per interaction)
  • Daily message subscriptions (₪49 per month)
  • Cross-promotion of digital kits, workshops, and consultations

This proposal defines the functional scope, technical architecture, and project estimations for both pricing options:

  • Fixed-Price implementation
  • Hourly model at ₪250/hour

The project will be developed in two incremental releases:

  • MVP (Minimal Viable Product) – predefined answers, menu automation, and payment links.
  • V1 (AI & Monetization) – personalized AI answers, payment verification, and subscription management.

Business Objectives

The project’s vision is to position Ilanit as a trusted digital mentor, combining intuitive empathy with structured automation.

Key objectives include:

  1. Human-like 24/7 Guidance
    Provide immediate emotional and motivational responses without religious or mystical overtones.

  2. Structured Life-Topic Support
    Cover common emotional challenges such as relationships, self-esteem, finances, career, and health through categorized entry points.

  3. Seamless Upsell Mechanisms
    Integrate natural CTAs leading to paid questions, subscriptions, or workshops — maintaining a non-salesy tone consistent with Ilanit’s brand.

  4. Automation & Scalability
    Use Azure serverless function, Logic Apps, or Rav-Meser automations to manage conversation flows, trigger AI responses, record logs, and route payments, enabling future scaling with minimal manual effort.

  5. AI Expansion Path
    Ensure the platform can later support AI assistants (ChatGPT/Azure OpenAI) that replicate Ilanit’s tone and philosophy through trained prompts and guardrails.

  6. Fast Go-to-Market Strategy
    Complete an MVP within ~4 weeks to validate the experience with real users, collect insights, and refine content before full-scale monetization.

Strategic Value

  • Strengthens Ilanit’s digital presence as a thought leader in emotional growth.
  • Creates a new recurring-revenue channel through WhatsApp — an already trusted, high-engagement platform.
  • Provides measurable analytics on user intent, engagement, and conversion from emotional triggers to paid actions.
  • Establishes a foundation for future AI-driven personal coaching and emotional-wellness products.

Technical Overview

The ILANIT Tarot platform is fully Azure-native, leveraging: - Azure AI Bot Service with Microsoft Bot Framework for omnichannel orchestration.
- Azure OpenAI Service for personalized, tone-controlled message generation.
- Azure Functions and Logic Apps for orchestration, payment integration, and automation.
- Azure SQL and Blob Storage for data persistence and content storage.
- Azure Key Vault for secrets, Application Insights for observability, and Power BI for analytics.
- Azure DevOps for CI/CD, backlog management, and code governance.

🌀 Functional Requirements

MVP (Core Release)

  1. WhatsApp Menu & Routing
    A structured WhatsApp experience will be delivered through the Microsoft Bot Framework, integrated with Azure AI Bot Service and connected to the WhatsApp Business channel (via Azure Communication Services or Rav-Meser).
    Users will be guided through four main emotional life areas:
    • Relationships & Love
    • Health & Wellbeing
    • Career & Money
    • Emotional Clarity / Personal Growth

Each selection triggers a predefined conversational flow handled by the Bot Dialog Manager, which interacts with Azure-based logic components.
The orchestration of the flow—intent recognition, message routing, and follow-up CTAs—is implemented via: 

  - **Azure Web Application** - custom developed microservices using **Microsoft Bot Framework** to realize bot logic.
  - **Azure Functions** – lightweight HTTP endpoints for message routing, tone validation, and dynamic menu generation.  
  - **Azure Logic Apps** – for optional low-code workflows (e.g., CRM integration, scheduling daily messages).  
  - **Azure SQL** – to log user sessions and menu interactions for analytics and bot conversations and data storage.

This architecture ensures high availability, scalability, and full telemetry through Azure Application Insights, while maintaining a consistent and warm conversational tone aligned with Ilanit’s brand.

  1. Content Repository
    Approximately 30 pre-written messages will be authored and stored within Azure-hosted services for secure and centralized management.
    The repository will reside in Azure Blob Storage as structured JSON or Markdown files, with optional synchronization from SharePoint/Drive or an internal CMS.
    Each content entry will include:
    • Topic category (Love, Health, Career, Personal Growth)
    • Message body (formatted text for WhatsApp display)
    • Signature line – closing statement with Ilanit’s tone of empathy and encouragement
    • Optional CTA – link to payment site, subscription, or external resource

Content will be loaded by the Bot Framework Dialog at runtime via an Azure Function API or direct Blob Storage SDK call.
Editors or non-technical staff will be able to update the repository through a secure content-management workflow (e.g., SharePoint or Logic App connector), ensuring real-time synchronization without code changes.
All access and updates will be protected by Azure Role-Based Access Control (RBAC) and Key Vault for connection secrets.

  1. Soft CTA Links
    Every response ends with a gentle invitation such as:
    “Would you like to go deeper? Click here to ask a personal question or receive daily guidance.”
    CTAs will redirect to landing pages for payments or subscription signups.
    These are implemented as short, trackable links for analytics and conversion tracking.

  2. Automation Layer
    The orchestration layer will be fully implemented within the Azure ecosystem, eliminating the need for third-party workflow tools.
    The Microsoft Bot Framework will handle conversational flow logic, supported by Azure Functions and Azure Logic Apps for background orchestration and integrations.

Key responsibilities of the automation layer include: - Message Intake & Routing: Incoming WhatsApp messages arrive via the Azure AI Bot Service (WhatsApp channel) and are processed by the Bot Application (.NET).
- Intent Recognition: The bot analyzes user input (menu selection or free text) and determines whether to use a predefined message or trigger an AI-generated response through Azure OpenAI Service.
- Function Triggers:
- Azure Functions (HTTP/Queue): Handle dynamic routing, AI inference requests, and background tasks.
- Azure Logic Apps (Optional): Execute longer-running or connector-based flows such as CRM updates, notifications, or scheduled message delivery.
- Telemetry & Logging: Each event (user ID, intent, timestamp, latency, and delivery outcome) is logged in Azure SQL Database and monitored through Azure Application Insights.
- Error Handling & Resilience: Built-in retries and alert rules detect failed message deliveries or AI timeouts; alerts are surfaced via Azure Monitor dashboards or email notifications.

  1. Basic Payment Flow
    Payment processing will be handled via a dedicated Azure-hosted payment microsite, integrated with a certified Israeli Payment Service Provider (PSP) such as Tranzila, CardCom, Pelecard, or Meshulam.
    This separation ensures PCI-compliant handling and secure tokenized transactions.

MVP Payment Flow: 1. The user selects “Ask a Personal Question” or another paid option in the WhatsApp menu.
2. The Bot Application generates a unique payment session link (e.g., https://payments.ilanit.co/session/{id}) hosted on the Azure payment site.
3. The user is redirected to the payment gateway for checkout (₪19).
4. Upon successful payment:
- The Israeli PSP sends a webhook notification to an Azure Function endpoint (/api/payments/webhook).
- The function updates the Azure SQL Database with payment status and associates it with the user session.
- The Bot Framework receives a signal to continue the conversation flow and sends a thank-you message confirming the payment.
5. Payment events and errors are tracked in Azure Application Insights for analytics and audit.

Security & Compliance Highlights: - All payment webhooks are HMAC-signed and verified in Azure Functions.
- Sensitive keys and tokens are stored in Azure Key Vault.
- The payment microsite runs on Azure App Service / Static Web Apps with HTTPS enforced and limited IP access.
- Payment status and history are accessible via Power BI dashboards built on Azure SQL data.

  1. Analytics & Logging
    All conversational, payment, and AI activity will be monitored and analyzed through Azure Application Insights, Azure Log Analytics, and Power BI dashboards.
    Each bot interaction or automation execution will record:
    • Topic chosen and intent classification
    • Message content and delivery status
    • CTA click events (e.g., “Ask a Personal Question,” “Subscribe”)
    • Response latency, AI inference time, and error codes

Logged data is stored in Azure SQL and Application Insights telemetry tables, then visualized via Power BI reports.
Core metrics include:
- Message volume and distribution by topic
- Engagement and CTA conversion rates
- Payment completion ratios
- Average response time and uptime statistics
- AI token usage and failure trends

Alerts and anomaly detection rules in Azure Monitor will proactively flag performance degradation or service errors.

  1. Pilot Launch
    A closed-group pilot (performed by customer) will validate the system before public release.
    The pilot will focus on:
    • Message clarity and alignment with Ilanit’s empathetic tone
    • Response time and perceived fluidity of the conversation flow
    • Technical reliability across the Bot Framework, Functions, and payment microsite
    • Data collection accuracy in Application Insights and Power BI dashboards

Feedback from testers/customer will be captured using structured forms or telemetry tags and analyzed.
After the pilot phase, adjustments will be applied to content, AI prompt tuning, and performance settings to prepare for production rollout.

V1 (AI & Monetization)

  1. AI-Personalized Responses
    Personalized responses will be powered by Azure OpenAI Service, integrated directly through Azure Functions or a secured Bot Framework skill endpoint.
    Key features:
    • Prompt Templates: Custom templates built in Azure Blob Storage and managed via versioned JSON configurations. Each template embeds Ilanit’s unique tone, vocabulary, and emotional guidance style.
    • Guardrails: Safety filters remove mystical, religious, or medical statements through Azure Content Filters and custom moderation policies.
    • Hybrid Mode: Option to deliver a predefined message introduction (from Blob) followed by an AI-generated continuation.
    • Contextual Awareness: The bot retrieves the user’s last interactions from Azure SQL to generate personalized, context-rich answers.
    • Telemetry: Each response includes metadata (tokens, latency, tone score) logged in Application Insights and available for review in Power BI analytics dashboards.
    • Fine-Tuning Readiness: Future phases may include fine-tuned Azure OpenAI models or embeddings to reflect Ilanit’s content corpus for more accurate and brand-consistent responses.
  1. Paid Question Flow (₪19)
    When a user completes a payment via the Azure-hosted Payment Microsite, integrated with an Israeli Payment Service Provider (PSP):
    • The Azure Function (Payment Webhook) validates and records the transaction in Azure SQL.
    • The bot transitions the user to a private conversation thread (session-scoped context) within the Bot Framework.
    • The user is prompted to submit a single personal question.
    • The AI Orchestration Function invokes Azure OpenAI to generate a warm, human-toned answer.
    • The reply is sent via the Bot Framework through the user’s active channel (WhatsApp, Web Chat, or Telegram).
    • The session auto-closes after response delivery, and a thank-you message is sent.
    • All session events (payment, question, response, latency, and token cost) are logged in Application Insights and stored in Azure SQL for future audits and reporting.
  1. Daily Message Subscription (₪49)
    The subscription system enables automated daily inspirational messages delivered through the user’s preferred channel.
    • Scheduler: Implemented using Azure Functions Timer Triggers or Logic Apps Recurrence workflows.
    • Message Source: Messages fetched from Azure Blob Storage (JSON/Markdown) and formatted via the Bot Framework Dialogs.
    • Personalization: Integration with Azure OpenAI (optional) to slightly vary tone or structure daily for repeat users.
    • Subscription Management:
      • Renewal reminders, pause, and cancellation managed through Logic Apps or Bot Adaptive Cards.
      • Payment renewals processed by the same Payment Microsite and tracked in Azure SQL.
    • Analytics: Message engagement rates, renewal success, and user retention visualized in Power BI.
  1. Human Handoff
    For sensitive or complex conversations, a human escalation path will be integrated using Azure Bot Framework Handoff Protocol or Logic Apps connectors.
    • Triggered when the AI moderation or sentiment analysis detects flagged keywords (e.g., emotional distress).
    • Conversation transcript and context are securely forwarded to Ilanit or an authorized human responder via a web-based admin panel hosted on Azure App Service.
    • Responses can be sent back through the bot channel, maintaining user continuity.
    • All interventions are logged in Azure SQL and traced in Application Insights for transparency and auditability.
  1. CRM & Marketing Integration
    Contact and engagement data will flow into Azure-based and third-party CRM systems through Azure Logic Apps and Data Connectors.
    • Contact Tagging: Automatically tag users as Free, Paid Question, or Subscriber within Azure SQL.
    • CRM Sync: Sync profiles and conversation metrics to connected systems (e.g., Rav-Meser, Dynamics 365, or Mailchimp) via Logic Apps connectors.
    • Analytics & Re-Engagement: Aggregate behavioral and financial data into Power BI for segmentation and targeted follow-ups.
    • Data Governance: All CRM integrations use secure connections, store minimal PII (hashed user IDs), and respect data retention and deletion policies configured in Azure Policy.

Non-Functional Requirements (NFR)

  • Tone & Branding
    • All conversational outputs must consistently express Ilanit’s warm, supportive, and non-religious personality.
    • Tone control is implemented through prompt templates and Azure OpenAI content filters.
    • Phrase filters and moderation logic in Azure Functions will remove or adjust prohibited or off-brand language before delivery.
    • All AI completions and prompt variations are logged in Azure Application Insights for tone-consistency audits.
  • Privacy & Security
    • All APIs, Bot endpoints, and webhooks are encrypted via HTTPS (TLS 1.2+) and authenticated using Azure Active Directory or Managed Identities.
    • Sensitive secrets and tokens (PSP keys, OpenAI credentials) are stored exclusively in Azure Key Vault.
    • User identifiers (phone numbers, chat handles) are hashed and anonymized in storage.
    • Data at rest is encrypted using Azure SQL TDE and Storage Service Encryption.
    • Data retention policies are enforced through Azure Policy and Lifecycle Management rules in Blob Storage.
    • Webhooks and external callbacks (e.g., payment confirmations) are HMAC-signed and verified within Azure Functions.
  • Performance
    • Average response latency target: ≤ 3–5 seconds for standard replies and ≤ 8 seconds for AI-generated responses.
    • The system must sustain 100–200 concurrent active chats during peak usage with autoscale enabled.
    • Azure Application Insights collects latency metrics and traces for performance analysis.
    • Azure Front Door or Traffic Manager may be used to optimize routing and reduce global response times.
  • Scalability & Resilience
    • Designed for elastic scale-out using Azure App Service autoscaling and Azure Functions Consumption/Premium plans.
    • Message queuing handled through Azure Service Bus or Storage Queue to buffer workloads during traffic spikes.
    • Stateless service design allows for horizontal scaling and zero-downtime deployment.
    • Future connectors (CRM, email, analytics) can be added through Azure Logic Apps or Event Grid without code changes.
    • AI calls include retry and circuit-breaker logic to ensure continuity during transient outages.
  • Observability & Operations
    • Unified monitoring and telemetry through Azure Monitor and Application Insights for the Bot App, Functions, and AI calls.
    • Centralized structured logging (JSON) with correlation IDs linking messages, payments, and AI responses.
    • Alerts & Dashboards:
      • Real-time error and latency alerts via Azure Monitor (email/Teams).
      • Power BI dashboards summarizing monthly usage, engagement, and SLA compliance.
    • Regular export of logs to Azure Log Analytics Workspace for long-term audit and compliance reporting.

Functional requirements for MVP and V1 are now defined and can be used to draft user stories, automation workflows, and acceptance criteria in the next planning phase.

Testing & Quality Assurance

  • Automated Unit Tests: Validate core services and AI integrations on every CI build.
  • Integration Tests: Verify payment flow, AI responses, and message routing across environments.
  • End-to-End Tests: Simulated WhatsApp sessions validate user flows pre-release.
  • Load Tests: Conducted via Azure Load Testing to ensure scalability for 100–200 concurrent sessions.
  • User Acceptance Tests (UAT): Executed during pilot to validate emotional tone and UX.

🌀 High-Level Architecture (HLD)

Stack principles:

  • Omnichannel via Microsoft Bot Framework + Azure AI Bot Service (no n8n/Zapier).
  • Orchestration on Azure Functions / Logic Apps only.
  • Storage 100% Azure (Blob/Queue/Table/SQL/Cosmos, Key Vault, App Insights).
  • Payments via a separate site that integrates with Israeli PSP (e.g., Tranzila / Pelecard / CardCom / Meshulam) using secure redirects + webhooks.
  • Components are first-party Azure wherever possible.

System Components

Layer Description
User Channels WhatsApp, Web Chat/Direct Line, Telegram, etc., federated via Azure AI Bot Service (Bot Framework).
Bot Application .NET (Bot Framework SDK) hosted on Azure App Service; dialog orchestration, state mgmt, guardrails, handoff triggers.
Azure Functions Stateless APIs and background tasks: AI inference proxy, content sync, payment webhooks, daily jobs.
Logic Apps (optional) Low-code connectors for CRM/email or long-running workflows (SLA escalations, retries).
AI Layer Azure OpenAI (or OpenAI) completions, prompt templates for Ilanit’s tone, moderation pipeline.
Content Store Authoring in SharePoint/Markdown/JSON; deployed to Azure Blob Storage with CDN; cached by the bot.
State & Data Azure SQL (users/sessions/payments audit), Azure Cosmos DB (optional for conversation transcripts), Azure Queue/Storage Table for jobs.
Payments Site Separate Azure App Service / Static Web Apps site with Israeli PSP integration. Redirect → PSP → PSP webhook → Azure Functions → Bot.
CRM & Analytics Azure Application Insights (telemetry), Power BI (dashboards), Logic Apps (email lists/CRM sync if needed).
Security Azure Key Vault, managed identities, signed webhooks, API Management (optional) in front of Bot/Functions.

Context Diagram (C4-Level 1)

flowchart LR
    U([User on Channels\nWhatsApp / Web / Telegram])
    ABS[Azure AI Bot Service\n - Bot Framework Channels]
    BOT[Bot App - .NET\nDialogs + Guardrails]
    FN[Azure Functions\nAI Proxy, Payments, Schedulers]
    AI[Azure OpenAI\nPrompted Tone]
    CONTENT[(Azure Blob Storage\nContent JSON/Markdown)]
    PAY[Payments Site\n - Azure App / Static Web]
    PSP[Israeli Payment Provider\n - redirect + webhook]
    SQL[(Azure SQL\nUsers/Sessions/Payments)]
    APPI[(Azure Application Insights)]
    KV[(Azure Key Vault)]
    LA[Logic Apps\n Optional workflows/CRM]

    U --> ABS --> BOT
    BOT --> FN
    FN --> AI
    BOT --> CONTENT
    BOT --> PAY
    PAY --> PSP --> FN
    BOT --> SQL
    FN --> SQL
    BOT --> APPI
    FN --> APPI
    BOT --> KV
    FN --> KV
    BOT <---> LA
Hold "Alt" / "Option" to enable pan & zoom

Deployment Topology (C4-Level 2)

graph TB
  subgraph Azure
    subgraph Channel
      ABS[Azure AI Bot Service\n - Bot Channels + Direct Line]
    end

    subgraph AppLayer[App Layer]
      BOT[Azure App Service\nBot Framework App - .NET 9]
      FN[Azure Functions\nHTTP/Webhook/Timers]
      LA[Logic Apps\n - Optional]
      APIM[Azure API Management\n - Optional]
    end

    subgraph DataAI[Data & AI]
      SQL[(Azure SQL Database)]
      COS[(Azure Cosmos DB - optional)]
      BLOB[(Azure Blob Storage + CDN)]
      Q[(Azure Queue/ Table Storage)]
      KV[(Azure Key Vault)]
      APPINS[(Azure Application Insights)]
      AOAI[Azure OpenAI]
    end

    subgraph Payments
      PSITE[Payments Site\n - App Service / Static Web Apps]
      PSP[Israeli PSP\n - redirect + webhook]
    end
  end

  ABS --> BOT
  APIM -. optional .-> BOT
  BOT --> FN
  BOT --> BLOB
  FN --> AOAI
  BOT --> SQL
  FN --> SQL
  BOT --> COS
  BOT --> Q
  BOT --> KV
  FN --> KV
  BOT --> APPINS
  FN --> APPINS
  BOT --> PSITE --> PSP --> FN
  LA -. CRM/Email Sync .-> FN
Hold "Alt" / "Option" to enable pan & zoom

Data Flow Overview

  1. Inbound: User → Channel (WhatsApp/Web/Telegram) → Azure AI Bot Service → Bot App.
  2. Routing: Bot analyzes intent (menu/topic vs. free text); either serves predefined content from Blob or calls AI via Functions.
  3. Response: Bot sends tone-aligned answer; soft CTA to Payments Site for paid Q or subscription.
  4. Payments: User completes payment at PSP; PSP webhook hits Functions → updates Azure SQL → Bot sends confirmation.
  5. Observability: Bot/Functions emit telemetry to App Insights; Power BI can visualize business KPIs.

Sequence – MVP (Predefined Answer via Bot)

sequenceDiagram
  participant U as User
  participant CH as Channel (ABS)
  participant B as Bot App (.NET)
  participant C as Content (Blob)
  participant S as SQL (Audit)
  participant I as App Insights

  U->>CH: Message / Menu selection
  CH->>B: Activity (Webhook)
  B->>C: Load predefined text + CTA
  C-->>B: Message payload
  B->>U: Reply (via Channel)
  B->>S: Log event (user/topic)
  B->>I: Telemetry (latency, success)
Hold "Alt" / "Option" to enable pan & zoom

Sequence – V1 (AI-Personalized + Payments)

sequenceDiagram
  participant U as User
  participant CH as Channel (ABS)
  participant B as Bot App (.NET)
  participant F as Azure Functions (AI/Payments)
  participant G as Guardrails (Policy)
  participant AOAI as Azure OpenAI
  participant P as Payments Site
  participant PSP as Israeli PSP
  participant S as Azure SQL
  participant I as App Insights

  U->>CH: Free text / Ask personal question
  CH->>B: Activity (Webhook)
  B->>B: Intent detect (paid flow?) 
  alt Not Paid
    B->>U: Send CTA + link to Payments Site
    U->>P: Redirect and pay
    P->>PSP: Payment request
    PSP->>F: Webhook (status=success)
    F->>S: Update payment record
    F-->>B: Notify bot (user authorized)
  end
  B->>G: Build prompt, enforce tone/policy
  G-->>B: Safe prompt
  B->>F: /ai/infer (prompt, context)
  F->>AOAI: Completion
  AOAI-->>F: Answer text
  F->>G: Post-check (safety/length)
  G-->>F: OK / adjust
  F-->>B: Final message
  B->>U: Personalized reply
  B->>S: Log (tokens, latency)
  B->>I: Telemetry + metrics
Hold "Alt" / "Option" to enable pan & zoom

Logical Data Model (Minimal)

erDiagram
  USER ||--o{ SESSION : has
  USER {
    string user_id "Channel user hash"
    string channel "whatsapp/web/telegram"
    string locale
    datetime created_at
    datetime last_seen
  }

  SESSION {
    string session_id
    string user_id
    string state "menu/awaiting_payment/paid_q/..."
    datetime started_at
    datetime updated_at
  }

  MESSAGE {
    string message_id
    string session_id
    string direction "in/out"
    string topic "love/health/career/emotions"
    int latency_ms
    datetime created_at
  }

  PAYMENT {
    string payment_id
    string user_id
    string type "paid_question/subscription"
    string provider "tranzila/pelecard/cardcom/..."
    string status "pending/success/failed"
    decimal amount
    datetime occurred_at
  }

  SUBSCRIPTION {
    string subscription_id
    string user_id
    string plan "daily_message"
    string status "active/canceled/paused"
    datetime started_at
    datetime renewed_at
  }

  SESSION ||--o{ MESSAGE : contains
  USER ||--o{ PAYMENT : makes
  USER ||--o{ SUBSCRIPTION : owns
Hold "Alt" / "Option" to enable pan & zoom

Bot Architecture (Dialogs & Policies)

  • Dialog stack:
    • RootDialog (welcome/menu/intent)
    • PredefinedDialog (static content → CTA)
    • PaidQuestionDialog (payment check → collect question → respond)
    • SubscriptionDialog (subscribe/unsubscribe/help)
    • HandoffDialog (route to human queue)
  • Policies/Guardrails: tone enforcement, banned phrase filters, sentiment checks, length control, rate-limit per user.

API & Webhook Contracts (Sketch)

# Bot incoming (from ABS to Bot)
POST /api/messages
Body: Bot Framework Activity (JSON)
Resp: 200 OK

# AI inference (Bot → Functions)
POST /ai/infer
Auth: AAD / Managed Identity / APIM
Body: { userId, locale, context:{topic?, lastN?}, inputText, mode:"personalized|hybrid" }
Resp: { text, safety:{filtered:boolean, reasons:[...]}, tokens:{prompt,completion,total}, latencyMs }

# Payment webhook (PSP → Functions)
POST /payments/webhook
Headers: X-Signature
Body: { paymentId, userRef, amount, currency, status, occurredAt }
Resp: 200 OK

# Bot notification (Functions → Bot)
POST /internal/payment/confirmed
Auth: Internal secret / AAD
Body: { userId, paymentId, type }
Resp: 200 OK

Observability & SRE

  • App Insights: end-to-end traces (ActivityId), custom metrics (p50/p95 latency, token usage, CTR, paid conversion).
  • Alerts: failed webhooks, AI timeout rate, spikes in handoff triggers.
  • Dashboards: Power BI reports (messages → CTA → payment → repeat usage).
  • Chaos/Resilience: retries/circuit breakers in Functions; transient fault handling in SDKs.

Security & Compliance

  • Key Vault for secrets (PSP keys, AOAI keys).
  • Managed Identity for Bot/Functions to access SQL/Blob.
  • Signed webhooks and HMAC verification for PSP + replay protection.
  • PII minimization: store channel IDs hashed; configurable retention (e.g., 90 days).
  • HTTPS-only, IP allowlists where applicable; optional APIM to centralize auth/rate-limits.

Scalability & Resilience

  • App Service autoscale for Bot; Functions Consumption/Premium for burst workloads.
  • Queue-backed jobs for scheduled messages and overflow processing.
  • Content caching in memory + Blob CDN to reduce latency.
  • Graceful degradation: fallback to predefined messages if AI is unavailable.

Technology Choices (initial)

  • Bot: .NET 9 + Bot Framework SDK, hosted on Azure App Service.
  • AI: Azure OpenAI (GPT family) with prompt templates & moderation.
  • Compute: Azure Functions for AI proxy, payment webhook, schedulers; Logic Apps for optional enterprise connectors.
  • Data: Azure SQL (core), Blob (content), Queue/Table (ops), Cosmos DB (optional transcripts).
  • Observability: Application Insights; Power BI for business dashboards.
  • Security: Key Vault, Managed Identities, APIM (optional).
  • Payments: Payments Site (App Service/Static Web) + Israeli PSP (redirect + webhook).

Open Decisions

  • Final WhatsApp channel approach (ABS with custom adapter vs. external bridge) within Azure-first constraints.
  • Choose the Israeli PSP and confirm webhook specs and testing sandbox.
  • Decide transcript store (SQL vs. Cosmos) based on volume and analytics needs.
  • Approve Power BI dataset model and refresh cadence.

Governance, Source Control & DevOps

All source code, infrastructure templates, and deployment pipelines are centrally managed in Azure DevOps, following secure, traceable, and auditable practices.

Source Control & Branching

  • The project uses a Git-based monorepo hosted in Azure Repos.
  • Branching strategy follows GitFlow:
    • main → production-ready releases.
    • develop → active sprint integration branch.
    • feature/* → individual features or experiments.
    • hotfix/* → urgent production fixes.
  • All pull requests require:
    • Code review approvals.
    • Automated build and test validation.
    • Security scan checks before merge.

Work Management & Sprints

  • Full lifecycle managed through Azure Boards:
    • Defined Epics, Features, User Stories, Tasks, and Bugs.
    • Each iteration follows 2-week sprints with sprint goals and velocity tracking.
    • Backlog grooming, daily standups, and retrospectives ensure continuous improvement.
  • Each work item is linked to corresponding commits, builds, and deployment records for full traceability.

CI/CD Pipelines

  • Continuous Integration and Continuous Deployment implemented with Azure Pipelines:
    • Build pipeline:
      • Restores dependencies, compiles code, runs unit and integration tests.
      • Executes static code analysis (SonarCloud / CodeQL) and security scans.
    • Release pipeline:
      • Deploys infrastructure and applications via Bicep or ARM templates.
      • Supports environment promotion: Dev → QA → Staging → Production.
      • Includes manual approval gates and automated smoke tests after deployment.
    • Pipelines secured using Service Connections with Managed Identities and Key Vault secrets.

Infrastructure as Code (IaC)

  • All Azure resources (App Services, Functions, SQL, Key Vault, Application Insights, Logic Apps) are provisioned through Bicep templates or Terraform modules.
  • Version-controlled IaC ensures reproducible, consistent, and auditable infrastructure environments.
  • Environment configuration (connection strings, API keys) sourced securely from Azure Key Vault.

Testing & Quality Gates

  • Automated unit tests (MSTest / xUnit) and integration tests run on each CI build.
  • End-to-end tests validate the Bot Framework conversation flow and AI responses via simulated sessions.
  • Azure Load Testing validates performance at peak concurrency (100–200 sessions).
  • Test coverage thresholds enforced; failing tests block merges or releases.

Deployment Security & Observability

  • All pipelines instrumented with Azure Monitor and Application Insights telemetry for deployment health.
  • Audit logs of pipeline runs, approvals, and configuration changes are retained for compliance.
  • Rollback and blue-green deployment strategies ensure zero downtime updates.
  • Security validations include:
    • Dependency vulnerability scans.
    • Secret detection and leakage prevention.
    • Role-based access control (RBAC) for all Azure DevOps projects.

Continuous Improvement & Reporting

  • DevOps metrics (deployment frequency, mean time to recovery, change failure rate) tracked through Azure DevOps Insights dashboards.
  • Operational metrics (latency, message throughput, AI cost per message) analyzed in Power BI.
  • Monthly reports summarize:
    • Sprint performance and delivered features.
    • System uptime and SLA compliance.
    • AI usage and cost efficiency trends.

This ensures that the ILANIT Tarot system is developed, tested, deployed, and operated within a fully governed, secure, and observable Azure DevOps ecosystem, supporting enterprise-grade scalability and long-term maintainability.

🌀 Project Delivery Phases, Estimations & Next Steps

Project Phases Overview

The ILANIT Tarot project will be implemented in structured, iterative phases to ensure quality, transparency, and incremental delivery.
All phases follow Azure DevOps-based governance, CI/CD automation, and sprint-based delivery.

Note: Future pricing estimations will exclude any external or third-party costs (e.g., Azure subscriptions, OpenAI usage, payment provider fees).

Phase 1 – MVP (Foundations & Predefined Experiences)

Objective: Deliver a functional WhatsApp bot experience using predefined content and the full Azure infrastructure foundation.

Key Deliverables: - Bot Framework app (.NET 9) deployed on Azure App Service with WhatsApp integration.
- Menu-driven conversational flow (Relationships, Health, Career, Emotions).
- Content repository on Azure Blob Storage with authoring workflow.
- Payment microsite (₪19 per question) integrated with Israeli PSP (sandbox).
- Logging and analytics pipeline using Application Insights + Power BI.
- Pilot environment configuration and test users onboarding.

Estimated Duration: 4–5 weeks (2 sprints)

Phase 2 – AI Personalization & Monetization (V1)

Objective: Introduce Azure OpenAI personalization, structured payment flow, and subscription automation.

Key Deliverables: - AI responses generated via Azure OpenAI with tone guardrails and moderation.
- AI orchestration through Azure Functions and prompt templates in Blob Storage.
- Secure webhook integration between Payment Microsite and Azure Functions.
- Subscription automation via Logic Apps or Timer Functions (₪49 daily message).
- Power BI dashboards for payment success rate, AI usage, and engagement metrics.
- Continuous deployment pipeline configured for QA and Staging environments.

Estimated Duration: 5–6 weeks (3 sprints)

Phase 3 – Human Handoff & CRM Integration

Objective: Enable smooth human escalation, CRM synchronization, and advanced analytics.

Key Deliverables: - Human handoff integration via Bot Framework Handoff Protocol and Logic Apps.
- Admin web panel on Azure App Service for manual message replies.
- CRM connectors (Rav-Meser, Dynamics 365, or Mailchimp) through Logic Apps.
- Extended telemetry (App Insights + Log Analytics) for full user journey tracking.
- Business intelligence datasets for retention and engagement in Power BI.

Estimated Duration: 4 weeks (2 sprints)

Phase 4 – Optimization, Scalability & Production Readiness

Objective: Hardening, observability, and full-scale deployment under production governance.

Key Deliverables: - Load testing using Azure Load Testing to validate 100–200 concurrent sessions.
- Performance optimization and latency tuning (AI and payment interactions).
- Deployment pipeline hardening (blue-green / canary strategy).
- Security validation: penetration tests, key rotation, policy enforcement.
- Final pilot feedback incorporation and production cutover.

Estimated Duration: 3–4 weeks (2 sprints)

Delivery Plan (High-Level Timeline)

Phase Key Focus Duration Major Deliverables
1 MVP – Core bot, menu, payment prototype 4–5 weeks WhatsApp integration, content flow, analytics
2 AI personalization & monetization 5–6 weeks Azure OpenAI, payment flow, subscriptions
3 Human handoff & CRM integration 4 weeks Admin panel, CRM sync, extended insights
4 Optimization & production readiness 3–4 weeks Load tests, scaling, observability

Total estimated duration: 16–19 weeks (8–10 sprints).
Includes parallel DevOps automation, testing, and content refinement.

Risks & Mitigation

Risk Description Mitigation
Payment integration delays Certification and API coordination with Israeli PSP may extend timelines. Engage PSP early, use sandbox first; isolate integration via Functions for flexibility.
OpenAI rate limits or response variability Azure OpenAI API quotas may impact performance or tone consistency. Pre-cache frequent topics, apply retries and fallback to predefined messages.
WhatsApp Business API constraints Message template approval or quota limitations from provider (360dialog / ACS / Rav-Meser). Register templates early, use approved categories; fallback to web or Telegram channels.
Performance degradation under scale Concurrency and latency spikes during campaigns or launch peaks. Autoscale App Service & Functions; enable Azure Front Door caching.
AI content sensitivity Responses may require extra moderation for tone or topic compliance. Implement content filters and human review escalation via HandoffDialog.
Cost management AI token usage and storage costs may rise with volume. Apply daily usage caps, Power BI tracking, and Azure cost alerts.

Dependencies

Dependency Description Responsibility
WhatsApp Business Account & Provider Connection setup via Azure Communication Services or Rav-Meser. Client + Dev Team
Israeli PSP Access Contract, API credentials, and sandbox keys for payment flow. Client
Azure Subscription & Resource Group Hosting, App Service, SQL, Key Vault, and Functions environment. Client (provisioned)
Content & Tone Guidelines Ilanit’s approved content library and language tone guide. Client
OpenAI Access Azure OpenAI resource approval and endpoint availability. Dev Team
Domain & SSL Certificates For payments.ilanit.** and bot endpoints. Client IT

Next Steps

  1. Finalize HLD & Document Sign-Off

    • Review and approve architecture and scope in this document.

  2. Azure Environment Setup

    • Provision resource group, app service plan, SQL, Blob, Key Vault, Application Insights.

  3. Project Initialization in Azure DevOps

    • Create repositories, pipelines, boards, and sprint plan.

  4. Content Preparation

    • Provide finalized 30 predefined messages and tone guidelines for MVP load.

  5. WhatsApp Provider & PSP Confirmation

    • Select and provision WhatsApp Business API and Israeli payment gateway sandbox.

  6. MVP Development Kickoff

    • Begin Phase 1 implementation (bot, content, payment prototype).

  7. Review & Iterate

    • Conduct sprint reviews, demos, and milestone validation via Azure DevOps Boards.

This phased delivery ensures controlled progress, transparent tracking, and reliable deployment within a fully Azure-native ecosystem, while keeping external service costs (Azure, OpenAI, PSP) out of development pricing estimates.

✅ Implementation Excellence & Partner Advantage

Partnering with Dmitry Khaymov (ConnectSoft) ensures that the ILANIT Tarot project benefits from proven, enterprise-grade foundations and the highest software engineering standards.

Why ConnectSoft & Dmitry Khaymov

  • Proven Azure-Native Template:
    The solution will be built using the ConnectSoft Cloud-Native Microservice Template, a mature, production-tested framework already implemented in multiple real-world systems.
    This template includes preconfigured modules for:

    • Azure Functions & App Service orchestration
    • Application Insights observability
    • Key Vault-based security and secret management
    • Automated CI/CD via Azure DevOps
    • API versioning, rate-limiting, validation, and telemetry
    • Integrated test automation and infrastructure-as-code

  • Cloud-Native & Fully Managed Design:
    Every component — from the Bot Framework to the AI and Payments microservices — follows Microsoft’s cloud-native best practices.
    The system is:

    • Serverless-first (scalable on demand, low operational overhead)
    • Fully managed by Azure (no manual hosting or infrastructure maintenance)
    • Secure by design with RBAC, encryption, and compliance built in
    • Observable end-to-end through unified monitoring, logging, and tracing

  • Accelerated Time-to-Market:
    Leveraging ConnectSoft’s ready-to-use architecture and DevOps pipeline templates, development starts from a stable, pre-validated foundation — reducing delivery time for MVP and ensuring predictable milestones.

  • Best-in-Class Engineering Practices:
    Dmitry and the ConnectSoft team apply consistent, world-class development methodologies:

    • Domain-Driven Design (DDD) & Clean Architecture principles
    • Automated testing (unit, integration, and acceptance layers)
    • Continuous Integration & Continuous Deployment via Azure Pipelines
    • Static analysis and code quality validation (SonarCloud / CodeQL)
    • GitFlow-based source control and peer-reviewed pull requests
    • Secure coding practices following OWASP & Microsoft SDL guidelines

  • Experience with Similar Systems:
    Dmitry’s previous delivery experience includes multiple Azure-based chatbot, AI, and automation solutions for enterprise clients.
    These implementations share architectural parallels with ILANIT Tarot — combining conversational AI, emotional guidance flows, and subscription-based digital services — ensuring rapid adaptation, proven reliability, and reduced project risk.

  • Quality, Flexibility & Long-Term Maintainability:
    The resulting platform will be:

    • Modular — easily extendable with new channels, content types, or AI services.
    • Compliant — adhering to Microsoft’s best practices for privacy and data governance.
    • Future-proof — ready for scale, analytics, and future AI integration (embeddings, personalization, etc.).
    • Enterprise-ready — auditable, monitored, and CI/CD-managed for continuous improvement.

Summary

By using ConnectSoft’s proven Azure-native solution stack and Dmitry Khaymov’s hands-on architectural expertise, the ILANIT Tarot system will be: - Technically robust and cloud-optimized
- Secure, scalable, and observable
- Delivered faster, with reduced development risk
- Maintained through automated, transparent DevOps processes

This approach ensures not only an efficient path to MVP and V1 but also a sustainable, high-quality foundation for future growth, integrations, and innovation — fully aligned with Microsoft’s modern cloud and AI standards.

💼 Commercial Models & Engagement Options

To provide flexibility and transparency, two engagement models are available — Fixed Price and Hourly Rate.
Both models follow the same engineering, quality, and DevOps standards already described in this document.

Note: Future pricing will not include external service costs such as Azure consumption, OpenAI usage, or Israeli payment provider transaction fees.

Purpose:
Best suited once the project scope, functionality, and integration boundaries are clearly defined.
This model ensures predictable delivery with milestone-based tracking and acceptance criteria.

Structure:
- Fixed cost per project phase (e.g., MVP, AI & Monetization, CRM Integration, Optimization).
- Includes all software engineering activities — architecture, implementation, QA, CI/CD automation, and documentation.
- Each milestone validated through Azure DevOps Boards, with deliverable visibility and acceptance gates.
- Scope changes managed via transparent Change Request (CR) process.

Advantages:
- Predictable timeline and budget.
- Minimal operational overhead for the client.
- Ideal for projects with well-defined functional boundaries.
- Enables early go-live and iterative post-launch optimization.

2. Hourly / Time-and-Materials Model

Purpose:
Recommended for projects where the functional scope may evolve or requires agile experimentation.
Allows iterative improvements and continuous collaboration while maintaining full visibility into effort and progress.

Structure:
- Hourly rate: ₪250/hour, covering development, architecture, testing, DevOps, and documentation.
- Progress and velocity tracked in Azure DevOps via sprint boards and automated reports.
- Weekly summaries with logged hours, tasks completed, and upcoming sprint goals.
- Flexible resource allocation — scaling up or down as priorities shift.

Advantages:
- Full flexibility for scope evolution and rapid iteration.
- Transparent cost tracking and time reporting.
- Suitable for integrations, enhancements, and post-launch support.

Engagement Governance (Applicable to Both Models)

  • All development, QA, and releases are managed in Azure DevOps with complete traceability.
  • CI/CD pipelines enforce build, test, and deployment validation automatically.
  • Each sprint concludes with:
    • A demo session presenting functional progress.
    • Retrospective review to adjust scope and efficiency.
    • Quality verification against acceptance criteria.
  • The client receives ongoing access to:
    • Source code repositories.
    • Infrastructure templates (Bicep/ARM).
    • Pipeline configurations and documentation.
    • Real-time dashboards for progress, quality, and deployment health.

Next Steps Toward Engagement

Once the preferred model and initial scope are confirmed, a formal Statement of Work (SOW) will be prepared, detailing: - Finalized scope and deliverables per phase.
- Timeline, sprint plan, and milestone checkpoints.
- Acceptance criteria and review process.
- Commercial terms and invoicing cadence.

This ensures the project is delivered under clear governance, proven Azure-native practices, and complete transparency — from architecture to production launch — guaranteeing both high quality and optimized time-to-market.

Perfect — here’s an additional subsection you can append right after the “Commercial Models & Engagement Options” section. It presents a transparent, approximate cost view for the Hourly (Time & Materials) model, using the delivery plan you already defined (Phases 1–4) and your rate of ₪250/hour.

It’s designed for professional client-facing use — clear, structured, and aligned with Azure project management language.

Estimated Workload & Effort Breakdown (Hourly Model – Approximation)

Below is an indicative breakdown of estimated effort per phase based on the delivery plan.
These estimations assume standard sprint velocity, Azure-native development practices, and full CI/CD automation.

Note: The following figures represent engineering effort only and exclude Azure consumption, OpenAI usage, and third-party service costs (e.g., PSP transactions).

Phase Scope Summary Estimated Effort (Hours) Approx. Cost (₪250/hr) Duration (Weeks)
Phase 1 – MVP Foundations Bot setup, predefined content flows, Azure infra, payment prototype, analytics pipeline 45–60 hrs ₪11,250 – ₪15,000 4–5 weeks
Phase 2 – AI Personalization & Monetization (V1) Azure OpenAI integration, tone guardrails, payment orchestration, subscription scheduler 55–70 hrs ₪13,750 – ₪17,500 5–6 weeks
Phase 3 – Human Handoff & CRM Integration Admin portal, human escalation, CRM sync (Logic Apps), extended telemetry 40–55 hrs ₪10,000 – ₪13,750 4 weeks
Phase 4 – Optimization & Production Readiness Load testing, observability, security hardening, blue-green deployment 30–45 hrs ₪7,500 – ₪11,250 3–4 weeks
Total (Approximate) End-to-end delivery across all phases 170–230 hrs ₪42,500 – ₪57,500 16–19 weeks total

Clarifications

  • These are initial, indicative estimates. The final effort will depend on:
    • Scope adjustments after client workshops.
    • Integration depth with existing systems.
    • Number of content items and complexity of workflows.
  • All work hours are logged and traceable via Azure DevOps Boards for full transparency.
  • Partial delivery or per-phase engagement is possible under the same hourly structure.
  • Additional iterations (post-pilot or scaling) can be added as separate mini-phases.

Example Delivery Cadence (Indicative)

Week Focus Area Main Deliverables
1–2 Environment setup, Bot baseline Azure infrastructure, initial flows, CI/CD pipeline
3–4 MVP menu + content + payments Full MVP tested on sandbox
5–8 AI & personalization rollout Azure OpenAI integration, tone tuning
9–12 Human handoff & CRM sync Admin panel, handoff dialog, Logic Apps
13–16 Optimization & production Load tests, observability, final deployment

This transparent estimation ensures that even under the hourly engagement model, the client retains full visibility into workload, costs, and milestones — maintaining the same rigor and accountability as a fixed-price project.

Fixed-Price Model – Phase-Based Structure (Indicative)

For clients who prefer a clearly defined cost and milestone-based delivery, the Fixed-Price Model provides full transparency with predictable scope, schedule, and deliverables.

All costs include architecture, development, testing, CI/CD automation, documentation, and project management —
and exclude any third-party or consumption-based expenses (Azure usage, OpenAI tokens, PSP fees, etc.).

Phase Description Duration (Weeks) Deliverables Pricing Model
Phase 1 – MVP Foundations Core bot setup (Azure AI Bot Service, WhatsApp integration, predefined flows, content repository, payment prototype, analytics) 4–5 Fully functional MVP with menu-driven WhatsApp experience, predefined answers, and payment microsite sandbox Fixed milestone price
Phase 2 – AI Personalization & Monetization (V1) Azure OpenAI integration, tone filtering guardrails, AI orchestration (Azure Functions), payment webhooks, subscription scheduler, dashboards 5–6 Personalized AI responses with moderation; live payment flow; daily message automation Fixed milestone price
Phase 3 – Human Handoff & CRM Integration Bot handoff dialog, admin portal, CRM connector (Logic Apps / Dynamics / Rav-Meser), extended analytics 4 Human escalation workflow + integrated CRM + advanced telemetry Fixed milestone price
Phase 4 – Optimization & Production Readiness Load testing, security hardening, observability improvements, CI/CD optimization, pilot feedback refinements 3–4 Production-ready system, monitored and optimized for reliability and scale Fixed milestone price

Engagement Structure

  • Each phase includes clear acceptance criteria, demo reviews, and client approval checkpoints.
  • Work progresses in sprints (2-week cycles) managed through Azure DevOps Boards, ensuring traceability and incremental delivery.
  • Upon completion of each milestone:
    • The client validates deliverables in staging.
    • A deployment sign-off is recorded in Azure DevOps.
    • The next phase begins upon formal approval.
  • Minor adjustments can be handled within a 10–15% flexibility window per phase without formal change requests.
  • Larger scope changes follow a Change Request (CR) process for approval and cost alignment.

Advantages of the Fixed-Price Model

  • Predictable Budget & Timeline: Each phase has a well-defined scope, cost, and delivery milestone.
  • Full Transparency: Progress, builds, and deployments visible via Azure DevOps dashboards.
  • Quality Assurance: Automated CI/CD pipelines ensure consistent build quality and deployment validation.
  • Low Risk: Delivery managed through formal acceptance and rollback procedures.
  • Strategic Alignment: The project roadmap follows Azure Well-Architected and AI Responsible Use frameworks.

Project Governance Summary

Activity Tool / Platform Owner
Backlog & Sprints Azure DevOps Boards ConnectSoft / Client
CI/CD Pipelines Azure Pipelines ConnectSoft
Source Code & IaC Azure Repos ConnectSoft
Testing & QA MSTest / Integration tests ConnectSoft
Monitoring & Logs Application Insights / Power BI ConnectSoft
Approvals & Change Control Azure DevOps (Change Requests) Client & ConnectSoft

Final Notes

  • The Fixed-Price engagement begins after the client and ConnectSoft finalize the detailed Statement of Work (SOW), which defines:

    • Functional scope and boundaries.
    • Milestones and acceptance criteria.
    • Deliverables and dependencies per phase.
    • Payment and invoicing milestones tied to acceptance.

  • This model guarantees:

    • High predictability in scope and schedule.
    • Strong governance through Azure DevOps.
    • End-to-end accountability with measurable outcomes.
    • Consistent quality through automated testing and validation.

Together, these ensure that ILANIT Tarot progresses from concept to production with clarity, control, and confidence — delivering a reliable, scalable Azure-native solution within a professionally managed framework.

💡 Engagement Models Overview

The ILANIT Tarot project can be delivered under one of two professional engagement frameworks —
both following the same high engineering standards, CI/CD automation, and Azure DevOps governance.

Criteria Fixed-Price Model Hourly (Time & Materials) Model
Best For Projects with a clearly defined and stable scope Projects with evolving scope or iterative experimentation
Scope Definition Agreed in advance and documented in the SOW Flexible and adaptable during development
Delivery Approach Milestone-based (per phase) with acceptance checkpoints Continuous sprints tracked via Azure DevOps Boards
Cost Structure Fixed per phase (predictable total budget) ₪250/hour, billed based on actual work hours
Budget Predictability High — predefined and locked per milestone Variable — depends on sprint velocity and change scope
Flexibility Limited to agreed CR buffer (10–15%) High — easy to add or adjust functionality
Governance Azure DevOps Boards + milestone reviews + acceptance criteria Azure DevOps sprint tracking + weekly time reports
Invoicing Per completed milestone and client acceptance Weekly or bi-weekly based on logged hours
Transparency Progress dashboards and milestone documentation Detailed time logs and sprint summaries
Change Management Through formal Change Request (CR) approval process Naturally handled through backlog reprioritization
Ideal Use Case MVP rollout and structured releases Research, AI tuning, post-launch optimization, integrations
Advantages Predictable cost, clear scope, low risk Flexibility, iterative feedback, adaptive priorities
Deliverable Visibility Milestone demos and sign-off checkpoints Continuous visibility in live DevOps dashboards
CI/CD & QA Fully automated build, test, and deploy pipelines Same CI/CD pipelines applied across all sprints
Azure Cost Inclusion Excludes Azure consumption, OpenAI usage, PSP fees Excludes Azure consumption, OpenAI usage, PSP fees
Ownership & IP 100% client-owned source code and documentation 100% client-owned source code and documentation

Model Selection Guidance

Both engagement models follow identical quality and security standards.
Choosing the appropriate model depends on whether the client prefers predictability (Fixed Price) or flexibility (Hourly).

ConnectSoft typically recommends beginning with the Fixed-Price MVP phase to establish the baseline system and then continuing with the Hourly model for further iterations and enhancements.