AI Smart Digital Procurement Platform in Kazakhstan
Explore a transformative digital procurement proposal for Kazakhstan using AI, blockchain, and machine learning to reduce corruption and increase efficiency.
Digital Public Administration (DPA 2502)
Astana, 2026
FINAL PROJECT
Smart Digital Procurement<br>Platform in Kazakhstan
Transforming Public Procurement Through AI & Digital Innovation
Maksim Moshko, Sabina, Altynay A.
[Instructor Name]
02
PROBLEM STATEMENT
Manual verification creates systemic latency and corruption risk.
THE PROBLEM
Current Procurement System Failures
goszakup.gov.kz relies on heavy manual verification → systematic delays, tender engineering abuse, and reactive (post-facto) auditing only.
PROJECT OBJECTIVES
Eliminate subjective human gatekeeping through algorithmic validation
Implement real-time ML screening for fraud & collusion detection
Build mathematical models to ensure system stability & scalability
03
ANALYTICAL FOUNDATION
E-Government 2.0 → 3.0
Once-Only Principle
GLOBAL BENCHMARKS
From Transactional to Cognitive Governance
EE
Estonia — X-Road
Decentralized data exchange layer. Absolute data integrity. Automated inter-agency trust via blockchain.
SG
Singapore — GeBIZ
Centralized identity (Singpass). Predictive market analysis. Automated risk profiling.
KAZAKHSTAN CONTEXT
High UN EGDI index — but back-end data silos prevent full automation.
04
SOLUTION DESIGN
Smart Digital Procurement Platform
AI document checking
Semantic anti-collusion screening
Decentralized ledger logs
TARGET USERS
Three-Tier Stakeholder Architecture
Procuring Entities
Ministry & akimat staff — Automated templates, market pricing recommendations
Private Suppliers
Micro, SME & large enterprises — Frictionless registration, instant qualification verification
Oversight Bodies
Anti-Corruption Agency & state auditors — Immutable logs, live analytics
05
USER FLOW
End-to-end procurement lifecycle with AI-integrated checkpoints
Initiation
Submission
Evaluation
Award
PROCESS ARCHITECTURE
AI-Integrated Procurement Flow
STEP 1 — INITIATION
Automated semantic scan of technical specs before publication — blocks restrictive criteria
STEP 2 — SUBMISSION
OCR/NLP processing of supplier credentials against state tax & legal databases
STEP 3 — EVALUATION
Algorithmic scoring + network fraud & IP collusion analysis
STEP 4 — AWARD
Immutable smart-contract execution — full transaction transparency
Each step is fully logged on distributed ledger
06
KPIs & METRICS
What gets measured, gets managed.
PERFORMANCE INDICATORS
Three-Dimensional Metrics Framework
07
FORECASTING
25-Year Vision
QUANTITATIVE PROJECTIONS
Phased Development Roadmap
3-Year Horizon
Stabilization
75% adoption in republic-level ministries <span style="color: #D88922; font-weight: bold; margin: 0 8px;">|</span> Cycle time: 12 days <span style="color: #D88922; font-weight: bold; margin: 0 8px;">|</span> Fiscal savings: $45M USD
5-Year Horizon
Universal Integration
100% adoption across regional akimats <span style="color: #C8973A; font-weight: bold; margin: 0 8px;">|</span> 90% automated processing <span style="color: #C8973A; font-weight: bold; margin: 0 8px;">|</span> Collusion reduced by 70%
10-Year Horizon
GovTech Maturity
Predictive economic instrument <span style="color: #1B2A4A; font-weight: bold; margin: 0 8px;">|</span> Joint procurement recommendations <span style="color: #1B2A4A; font-weight: bold; margin: 0 8px;">|</span> Cumulative savings: $320M USD
25-Year Horizon
Cognitive Governance
Fully autonomous infrastructure <span style="color: #2C2C2C; font-weight: bold; margin: 0 8px;">|</span> Self-generating smart-contract micro-tenders <span style="color: #2C2C2C; font-weight: bold; margin: 0 8px;">|</span> 99.9% uptime
08
OPTIMIZATION
-40%
Time Saved
-30%
Cost Saved
100%
Proactive Audit
EFFICIENCY GAINS
From Reactive Control to Inline Algorithmic Shield
40% reduction in administrative latency through real-time API integrations
24 days → 5–12 days
30% reduction in average administrative transaction costs per tender
Direct fiscal efficiency
Moving from post-facto cameral control to inline algorithmic risk flagging
Pre-emptive, not reactive
BEFORE: Post-facto audit
AFTER: Real-time inline shield
RESOURCE ALLOCATION
Institutional Resource Matrix
Resource Category
Asset Description
Core Institutional Role
👤 Human
Senior ML Engineers, Blockchain Architects, GovTech Security Specialists
Algorithmic screening, immutable logging, zero-knowledge security
💰 Financial
Core Development, Continuous Maintenance, Public Training Budget
Backend software, API integration, regional onboarding
🖥️ Technological
Distributed Cloud Clusters, Cross-Agency API Gateway, TLS 1.3 / AES-256
High-availability hosting, data interoperability, data protection
10
RISK ANALYSIS
P(Alert) = Σ[P(Rᵢ) × P(Alert|Rᵢ)]
+ [P(No Risk) × P(Alert|No Risk)]
= 0.4205
Total Alert Probability
BAYESIAN INFERENCE
Dynamic Risk Probability Assessment
Bayesian inference allows public managers to dynamically calculate the true probability of a threat given an automated platform anomaly alert.
R₁
Technical Failure
Prior: 15%
R₂
Cyber Attack
Prior: 8%
R₃
Low Adoption
Prior: 25%
R₄
Algorithmic Bias
Prior: 10%
Each risk probability is updated dynamically upon system alert detection
11
BAYESIAN RESULTS
An automated alert is most likely to indicate infrastructure failure or cyber breach (51.01% combined)
→ Prioritize technical redundancy
PRIOR vs. POSTERIOR PROBABILITIES
Updated Risk Distribution After Alert
R₁ — Technical Failure
15%
33.89%
↑ 18.89 pp increase
R₂ — Cyber Attack
8%
17.12%
↑ 9.12 pp increase
R₃ — Low Adoption
25%
23.78%
↓ slightly reduced
R₄ — Algorithmic Bias
10%
20.21%
↑ 10.21 pp increase
R₁ + R₂ Combined Posterior: 51.01% → Shift resource priority to infrastructure resilience
12
DEMAND MODELING
Bass Diffusion Model
35,000
Public Entities
180,000
Private Suppliers
Total Addressable Pool
ADOPTION KINETICS — BASS DIFFUSION MODEL
Three-Scenario Demand Forecasting
BASELINE SCENARIO
+8% annual growth
Stable growth — 20,000 concurrent peak requests/hour
Handles routine seasonal budget cycles
OPTIMISTIC SCENARIO
+18% annual growth
Mandatory integration of Samruk-Kazyna holdings — 45,000 peak requests/hour
Requires automated cloud sharding & Kubernetes orchestration
PESSIMISTIC SCENARIO
+2% annual growth
Institutional stagnation in rural areas
Requires regional IT expansion & interface simplification
Model calibrated to Kazakhstan's eGov adoption trajectory (2020–2025 baseline)
13
CONCLUSION
Thank You
Smart Digital Procurement Platform in Kazakhstan
Digital Public Administration (DPA 2502)
Astana, 2026
SUMMARY & FEASIBILITY
A Transformative, Viable, and Scalable Solution
VALUE PROPOSITION
Maximizes public value through transparency, immutability, and projected <strong style="color: #4A4A4A; font-weight: 600;">12–15% reduction</strong> in state procurement expenditure.
FEASIBILITY IN KAZAKHSTAN
Highly viable — existing <strong style="color: #C8973A; font-weight: 600;">eGov</strong> infrastructure and digital signature (EDS) ecosystem provide strong foundation for rapid deployment.
SUCCESS PREREQUISITES
Requires: strong political will for database integration, legislative structural updates, and proactive civil service digital training.
Questions & Discussion Welcome
- govtech
- digital-procurement
- kazakhstan
- e-government
- ai-in-government
- blockchain
- public-administration