Durban, KwaZulu-Natal — Procurement Intelligence Platform

South Africa's procurement
processes deserve
better.

CivicFlow AI automates supplier evaluation across 8 weighted dimensions — producing AI-assisted procurement scoring that is fully explainable, legally defensible, BBBEE-aligned, and designed to create the documentation that protects officials and withstands scrutiny.

Request a Pilot Analysis See how it works
8
Scoring Dimensions
25%
BBBEE & Local Impact Weight
100%
Explainable Outputs
Open
Public Methodology v1.0
R268bn
Municipal irregular, fruitless & wasteful expenditure
National Treasury MFMA Compliance Report, 2024/25
80%
Of irregular expenditure cases left unresolved nationally
Auditor-General South Africa, 2024
4
KZN municipalities with clean audits in 2022/23, down from 9
AGSA Local Government Audit Outcomes, 2022/23
The Problem

Inadequate documentation is driving irregular expenditure — not just corruption.

National Treasury has explicitly noted that irregular expenditure can arise from inadequate documentation even where spending itself was legitimate. This is the problem CivicFlow AI solves: not catching criminals, but creating the structured documentation and evidenced scoring that protects both public funds and the officials responsible for them.

RFQ procurement under R500,000 is the highest-volume, lowest-scrutiny layer of South African public spending. Quote selection is discretionary, scoring is inconsistent, and audit trails are often inadequate — creating findings that follow municipalities for years.

CivicFlow AI automates the analysis layer — producing structured, evidenced, explainable scoring that creates a defensible record on every procurement event, every time.

R500k
The RFQ threshold — South Africa's highest-volume procurement tier, where discretionary decision-making is most common and documentation most often inadequate
§217
The constitutional clause requiring all public procurement to be fair, equitable, transparent, competitive, and cost-effective — CivicFlow AI makes each of these measurable and documentable
Zero
Black-box outputs — every CivicFlow score references specific source data, states its confidence level, and discloses every data gap and limitation
How It Works

Three steps from raw data
to defensible recommendation.

01
Data Collection & Verification
Supplier data is gathered from lawful public sources and submitted documentation. Every data point is timestamped and sourced. Where information is unavailable or unverifiable, this is recorded explicitly — never assumed or estimated without disclosure.
National Treasury CSD SARS PIN Verification CIPC Directorship Records BBBEE Commission PAIA Requests AGSA Findings Database
02
AI-Assisted Scoring Across 8 Dimensions
Each supplier is evaluated across 8 weighted dimensions covering price competitiveness, transformation, technical capability, past performance, financial health, compliance, local impact, and statistical anomaly detection. Every sub-score is evidenced. A composite score out of 100 is produced with a full confidence rating.
BBBEE Weighted at 20% Local Impact at 5% Anomaly Detection Confidence Rated Comparative Narrative
03
Explainable Report & Audit Trail
Output is a structured report with dimension-level breakdowns, evidence citations, confidence levels, anomaly flags, and a comparative narrative across all evaluated suppliers. Every output is designed to be read by a procurement official, presented to a council, or reviewed by an auditor — without modification.
Full Evidence Citations Confidence Level Stated Anomaly Flags Audit Ready Version Controlled
Scoring Methodology v1.0 — Public Document

Eight weighted dimensions.
Every score explained.

The full methodology is publicly available. Every weighting decision is documented and justified. Scores are estimates based on available evidence — never presented as objective truth. Confidence levels are always stated. Missing data is always disclosed.

01
Price Competitiveness
20%
02
BBBEE Compliance & Transformation
20%
03
Technical Capability & Experience
20%
04
Past Performance Record
15%
05
Financial Health & Stability
10%
06
Compliance & Registration Status
8%
07
Local & Community Impact
5%
08
Anomaly & Risk Indicators
2%

Transformation-first design: BBBEE Compliance (20%) and Local & Community Impact (5%) together represent 25% of the composite score — directly reflecting the constitutional mandate of Section 217(2) and the Preferential Procurement Policy Framework Act. BBBEE scoring is a core feature, not an afterthought.

Who It's For

Three types of client.
One platform.

CivicFlow AI serves any organisation that needs procurement to be faster, more defensible, or more transparent — regardless of size or sector.

🏛
Municipalities & Government SCM Teams
Automate RFQ supplier evaluation, generate defensible scoring records, and reduce the risk of AGSA irregular expenditure findings caused by poor documentation. Build audit trails that protect officials and withstand public scrutiny — without adding headcount.
Currently accepting pilot engagements in KwaZulu-Natal. eThekwini municipality procurement data is CivicFlow AI's primary reference dataset.
⚖
Civil Society, NGOs & Opposition Offices
Commission independent procurement analysis on specific tenders or spending categories. Receive statistically grounded findings — framed as pattern reporting, not accusations — that are suitable for advocacy, media engagement, or formal complaints without creating legal exposure.
Watchdog organisations, investigative journalism units, political accountability offices, civil society coalitions.
🏢
Private Sector & SOE Procurement Teams
Apply structured AI-assisted scoring to internal procurement. Reduce supplier risk, enforce transformation targets, automate evaluation workflows, and create documentation boards, auditors, and B-BBEE verification agencies can rely on.
Corporates, state-owned enterprises, healthcare groups, construction companies, professional services firms.
Services & Pricing

From single audit to full platform.

Every engagement tier delivers standalone value. Clients typically start with a Process Audit Report and move up the ladder as confidence in the system grows.

Tier 1 — Entry Point
Process Audit Report
R5,000 – R25,000
Independent analysis of a defined procurement event or category. Scored across all 8 dimensions, fully evidenced, and delivered as a formal report with confidence ratings, anomaly flags, and a comparative supplier narrative.
Ideal for: NGOs, civil society, opposition offices, municipalities wanting a once-off review before committing to automation
Tier 2 — Automation
Workflow Automation Build
R30,000 – R120,000
Custom automation of a specific procurement workflow — RFQ scoring, supplier shortlisting, or anomaly flagging — built, tested, and deployed for your organisation's exact process and data environment.
Ideal for: Municipalities and corporates with recurring procurement cycles wanting to eliminate manual evaluation
Tier 3 — Transparency
Transparency Dashboard
R80,000 – R350,000
Full procurement transparency dashboard with live supplier scores, anomaly flags, confidence ratings, and complete audit trails. Configurable for internal management use, council reporting, or public accountability publication.
Ideal for: Municipalities, SOEs, and organisations with public accountability obligations or council oversight requirements
Tier 4 — Platform
SaaS Platform Access
R8,000 – R45,000/mo
Ongoing platform access with automated scoring, supplier database management, real-time anomaly monitoring, and versioned reporting. Fully managed infrastructure, continuously updated data, fully explainable outputs.
Ideal for: Organisations with continuous, high-volume procurement needs requiring ongoing intelligence rather than once-off analysis
Why Trust CivicFlow AI

Built for legal defensibility
from day one.

⚖
Legally Conservative by Design
All outputs are framed as statistical analysis and pattern reporting — never legal findings or accusations. CivicFlow AI never makes binding procurement decisions, never draws legal conclusions, and never names individuals in automated public reports. Anomaly flags are statistical observations requiring human review, not verdicts. This framing is not a disclaimer — it is the core design principle.
📋
Open, Versioned Methodology
The full scoring methodology is publicly available as a versioned document. Every weighting decision, every data source, every limitation is stated explicitly. Version 1.0 was published May 2026 and is formally reviewed every 6 months. Previous versions are archived and remain accessible. Stakeholder feedback from municipalities, civil society, and suppliers is formally considered in each review cycle.
🔍
Confidence Levels Always Stated
Every recommendation carries an explicit confidence rating — HIGH (85–100%), MEDIUM (60–84%), or LOW (below 60%) — based on data completeness. Missing data is always disclosed in outputs. Scores are estimates based on available evidence, never presented as objective truth. A LOW confidence output is published as LOW — it is never suppressed or inflated.
🏛
Aligned to South African Law
Designed in alignment with the Constitution (§217), MFMA, PPPFA, B-BBEE Act, PRECCA, and PAIA. Data is collected only through lawful means — public records, PAIA requests, and supplier-submitted documentation. No private or confidential information is used without legal authorisation or explicit consent.
🌍
Transformation-First Scoring
BBBEE Compliance and Local & Community Impact together represent 25% of the composite score — making transformation a structural priority, not a compliance checkbox. The methodology is deliberately designed so that a highly competitive price cannot override a supplier's transformation credentials.
📊
Data Sourced, Timestamped, Refreshed
Every data point records when it was collected and from where. Supplier data is refreshed at each procurement event — historical scores do not automatically carry forward. The system does not make assumptions about data that cannot be verified: it flags the gap and adjusts the confidence level accordingly.
System Boundaries

What CivicFlow AI will never claim.

The system reduces opacity and improves documentation. It does not replace human judgment, and it does not generate legal liability based on automated outputs alone. These boundaries are permanent and non-negotiable.

Will never state that a supplier is corrupt or has committed fraud
Will never make a binding procurement decision — recommendations and analysis only
Will never present a score as objective truth — all scores are evidence-based estimates
Will never omit limitations or missing data from its outputs
Will never use personal data beyond legally available public procurement records
Will never produce outputs without confidence levels and evidence citations
Will never be positioned as a replacement for qualified SCM officials
Confidence Framework

Every output carries an explicit confidence rating based on data completeness. Low-confidence outputs are flagged — never suppressed, never inflated.

LevelRangeMeaning
HIGH85–100%All 8 dimensions verified. Recommendation is fully evidence-backed and audit-ready.
MEDIUM60–84%Most dimensions verified. 1–2 estimated or partially evidenced. Gaps disclosed.
LOWBelow 60%Significant data gaps exist. Treat as indicative only — additional verification required before acting.
Primary Data Sources
National Treasury CSD  ·  SARS PIN Verification  ·  CIPC Directorship Records  ·  B-BBEE Commission  ·  Auditor-General SA  ·  Municipal Procurement Portals  ·  PAIA Requests  ·  Supplier-Submitted Documentation
Frequently Asked Questions

Questions procurement professionals
ask before engaging.

Yes. CivicFlow AI outputs are framed as statistical analysis and advisory recommendations — not administrative decisions. The system produces intelligence that informs procurement officials; it does not make binding decisions. All final procurement decisions remain the legal responsibility of authorised SCM officials in accordance with the MFMA, PPPFA, and applicable supply chain regulations. The methodology has been deliberately designed to be legally conservative from the outset, and formal legal review of all outputs is recommended before first public use.
Anomaly flags are statistical observations — not findings of wrongdoing. The system is explicitly designed to surface patterns that warrant documented human review, not to draw conclusions. Many anomaly flags will have entirely legitimate explanations. The output directs attention and requires documentation of the explanation — which is exactly what good procurement practice requires. A flag is not a finding. It is a prompt for the kind of documented justification that protects officials.
No. CivicFlow AI's primary data sources are public records, published tender notices, award announcements, and formal PAIA requests. Municipalities are not required to cooperate for independent analysis to be conducted. When municipalities engage directly as clients, they can provide additional internal data that improves scoring confidence — but this is not a prerequisite. Civil society clients and NGOs can commission analysis independently.
BBBEE Compliance and Local & Community Impact together represent 25% of the composite score — making transformation a structural scoring priority, not a compliance footnote. The weighting aligns directly with the Preferential Procurement Policy Framework Act and the constitutional mandate of Section 217(2). The scoring methodology rewards higher BBBEE levels and local economic participation. A supplier with superior transformation credentials will outperform a competitor on these dimensions regardless of price. The system is designed to support transformation goals, not constrain them.
Process Audit Reports start at R5,000. This tier is specifically designed for organisations that want to test the system on a defined procurement event before committing to ongoing automation. A single report covering one RFQ category is a practical starting point that delivers immediate value — a scored, evidenced, audit-ready supplier evaluation — without requiring a technology budget. Pilot pricing is available for first engagements in KwaZulu-Natal.
CivicFlow AI collects only data that is legally available in public procurement records or explicitly submitted by the engaging organisation. No private or confidential supplier information is used without legal authorisation. Data is timestamped and sourced. Supplier data is refreshed at each procurement event — historical data does not automatically carry forward into new scoring cycles. Full data handling terms are provided at the point of engagement.
Get Started

Ready to bring structure
to your procurement process?

CivicFlow AI is currently accepting pilot engagements in KwaZulu-Natal. Process Audit Reports are available now. Contact us to discuss your specific procurement challenge — we respond to all enquiries within 48 hours.

R5,000
Pilot audit starting price
KZN
Current pilot focus area
48hrs
Response to all enquiries
hello@civicflowai.co.za

civicflowai.co.za  ·  Durban, KwaZulu-Natal, South Africa