Soudah AI — Finance
Multi-Agent Intelligence Platform
Start small. Prove value. Scale with confidence.
Agents at Phase 1
Agents at Full Scale
Cloud Go-Live
On-Prem Go-Live
Prepared by Zoé AI Solutions — Saudi Arabia & GCC • Version 2.0 (2025)
Three Critical Challenges
Financial operations face systemic inefficiencies requiring highly measurable AI interventions
Inefficiencies in Accounts Payable (AP) due to multiformat invoice submissions across unmanaged channels, manual classification, validation, and routing processes.
Multi-Agent Ecosystem
A coordinated network of specialized AI agents working together to transform financial operations
Orchestration Agent
Coordination
Routes queries, sequences activities, central system orchestrator.
Document Intelligence Engine
Document Processing
Extracts, classifies, and validates invoice data from multiformat documents.
Anomaly Detection Agent
Statistical Outlier Detection
Identifies statistical anomalies against historical norms.
Fraud Detection Agent
Pre-Payment Flagging
Applies fraud guardrails before payments. ML model matures in Phase 2.
Predictive Analysis Agent
Forecasting
Budget, cash flow, and volume projections.
Behavioural Analysis Agent
Pattern Context
Establishes behaviour baselines for departments/vendors.
Sentiment Analysis Agent
Communication Risk
Evaluates risk signals from communications.
Prescriptive Analysis Agent
Recommendations
Generates and prioritizes actionable financial recommendations.
How the Agents Work
A cascading workflow that processes, analyzes, and flags financial transactions in real-time
Phase 1 Cascade — 4 Steps
Orchestrator
Invoice data received and routed for analysis
Anomaly Detection
Reviews outliers against historical norms
Fraud Detection
Fraud rules applied, holds flagged if thresholds met
Dashboard
Dashboard updates with alerts and actions
Orchestrator
Invoice data received and routed for analysis
Anomaly Detection
Reviews outliers against historical norms
Fraud Detection
Fraud rules applied, holds flagged if thresholds met
Dashboard
Dashboard updates with alerts and actions
Three-Phase Delivery
Each phase builds on a functioning system, transitioning based on measurable performance
Core MVP
Orchestration + Document Intelligence + Anomaly & Fraud Detection
Click to see deliverables
Intelligence Augmentation
Predictive Modelling + Zoé Spark Onboarding
Click to see deliverables
Full Ecosystem
7-Agent Model — Fully Integrated Alerts, Sentiment Checks & CFO Advisories
Click to see deliverables
Measurable KPIs
Each phase transition is gated by specific, measurable performance criteria
Invoice Cycle Reduction
Reduction vs. baseline processing time
Data Extraction Accuracy
200 randomly sampled documents
Fraud Alert Precision
Genuine alerts (rule-based; improves in Phase 2)
Why 60% Precision in Phase 1? Acknowledges the limitations of rule-based systems. Errors surfacing in Phase 1 are intended for correction before moving to ML-based detection in Phase 2.
Phase 1 → Phase 2 — Target KPIs
Zoé Spark
Data Scientist as a Service
Empowering finance teams with natural language access to ERP data, ML predictions, and automated dashboards
Multi-Source Connection & Schema Inference
Auto-maps ERP connections for real-time integration within a single 4-hour review session.
Natural Language Querying
Ask questions in plain language and get instant results with charts and tables.
On-Demand ML Modelling
Triggered via natural prompts for spend predictions, vendor analysis, and forecasts.
BI Dashboard & Report Generation
Generates shareable, automated dashboards and reports in English or Arabic.
In-Scope vs. Out-of-Scope
Clear boundaries ensure focused delivery and measurable outcomes at each phase
In Scope
- Deploy Document Intelligence Engine
- Configure Anomaly & Fraud Detection Agents
- AP Dashboard setup
- Conduct rule-based training
Out of Scope
- Hardware procurement
- ERP licensing/configuration
- Multi-ERP system support
- Non-Arabic/English documents
- Legal/compliance advisory
Choose Your Deployment
Cloud, On-Premises, or Hybrid — the client's choice based on security, compliance, and speed requirements
Cloud Deployment
- Zoé AI handles all cloud resources
- AWS Bahrain or Azure UAE
- NDMO-compliant data residency
- Fastest time-to-value
AWS Bahrain or Azure UAE (NDMO-compliant)
NDMO-compliant data residency in KSA region
6 weeks to Go-Live
Transparent Pricing Structure
Contract specifies a maximum price ceiling for each phase. Final itemized pricing is confirmed after scoping.
Client Protection
The Discovery Workshop (Days 1–2) validates volumes, ERP complexity, and data quality. Final itemized price issued within 3 business days. Clients have the right to terminate if Zoé cannot price Phase 1 within the ceiling.
Next Steps
A clear roadmap from proposal acceptance to Go-Live
Select Deployment Model
Cloud, on-premises, or hybrid. Decision within 5 days post-proposal acceptance.
Appoint Sponsors
Executive Sponsor (CFO/Deputy) and Programme Manager (50% time) within 5 days.
Hardware Procurement
For on-premises: begin parallel to signing. Must be racked by Week 4.
Contract Signing
Contract including Discovery Workshop finalized within 15 days.
Discovery Workshop
Two-day onsite session with Finance Ops Lead, IT Lead, Internal Audit Lead, CFO Representative, and Zoé Solutions Architect.
ERP Historical Data Extraction
IT teams extract 12–24 months of historical data for ML training.