SenATIQ Conseil Logo
← Home
Thought Leadership & Insights

Technical and strategic analysis on data sovereignty, regulatory compliance, and localized AI in West Africa.

Data SovereigntyJuly 2026·8 min

Data Residency in Dakar: PAIX vs SENUM SA — Which Datacenter for Your Organization?

With the National Data Strategy (2023–2028) in force and rising digital sovereignty requirements, Senegalese organizations face a strategic choice: host critical data with a private operator like PAIX or in the national datacenter operated by SENUM SA.

The Dakar datacenter landscape

Three players structure data hosting in Senegal today. PAIX Data Centres, backed by the pan-African fund Africa50, operates a carrier-neutral datacenter in Dakar and is a major Internet exchange point for the sub-region. Sénégal Numérique S.A. (SENUM) operates the Diamniadio National Datacenter, dedicated to hosting public and administrative data. Finally, Onix Data Centres, located in Les Almadies, hosts the 2Africa submarine cable landing station.

The latency imperative

Local hosting reduces network latency to 5–12 milliseconds, versus 65–90 milliseconds for European cloud hubs. For real-time banking applications, Open Banking APIs, and event-driven architectures, this gap is decisive for user experience and compliance with BCEAO real-time reporting requirements.

The regulatory framework

Law No. 2008-12 on personal data protection and the National Data Strategy 2023–2028 push toward local residency for sensitive and public data. The CDP (Commission for the Protection of Personal Data) regulates cross-border transfers, making local hosting increasingly relevant for regulated sectors.

Decision criteria

Choosing between PAIX and SENUM SA depends on several factors: the nature of the data (public vs private), sovereignty requirements, Internet exchange connectivity needs, certification (ISO 27001, Tier III/IV), and the level of redundancy required. A private operator generally offers more flexibility and managed services, while the national datacenter guarantees enhanced sovereignty for public data.

BCEAO RegulationJune 2026·12 min

BCEAO Instruction 2024: API Architecture Standards for UEMOA Payment Institutions

BCEAO Instruction No. 001-01-2024 redefines the payment services framework within UMOA and imposes precise technical requirements on payment institutions in the UEMOA zone. Here are the concrete architectural implications.

Scope of the instruction

Instruction No. 001-01-2024, in force since 2024, governs payment services in UMOA member states. It is part of the Open Banking dynamic driven by the BCEAO and requires payment institutions to modernize their information systems, notably through the exposure of standardized APIs.

Architectural implications

Concretely, payment institutions must deploy: a mandatory API Gateway to expose and secure their services; OAuth 2.0-based authentication and authorization mechanisms; TLS 1.3 encryption for all exchanges; systematic transaction logging; and real-time reporting obligations to supervisory systems.

Compliance challenges

For banks and fintechs in the UEMOA zone, compliance often implies reworking the existing architecture: breaking down legacy monoliths, establishing an API-first foundation, and integrating resilience and traceability constraints. The difficulty lies less in technology than in the architectural governance needed to align existing systems with these new requirements.

A progressive approach

Our recommendation is a Strangler Fig approach: progressively expose compliant APIs around the existing system, without service interruption, prioritizing critical regulatory use cases. This approach delivers value quickly while controlling operational risk.

AI & Wolof NLPMay 2026·10 min

Deploying Wolof Voice AI in Retail Banking: Technical Challenges and Strategic Opportunities

Wolof is spoken by over 80% of Senegal's population but remains largely underrepresented in the training corpora of large language models (LLMs). Deploying Wolof voice AI is both a technical challenge and a major strategic opportunity for financial inclusion.

An underrepresented language market

Global LLMs are predominantly trained on English and, to a lesser extent, French corpora. Wolof, Senegal's national lingua franca spoken by the vast majority of the population, remains absent from major public corpora. This underrepresentation limits the quality of voice and conversational services in local languages.

Technical challenges

Three major challenges arise: collecting annotated Wolof voice data (costly and poorly standardized), fine-tuning speech recognition (ASR) and synthesis (TTS) models on African corpora, and evaluating production quality amid dialectal diversity. Integrating Wolof–French code-switching, common in everyday speech, adds further complexity.

Banking use cases

For retail banking, Wolof voice AI opens high-impact use cases: local-language customer assistance, voice onboarding for low-literacy populations, and conversational agents for basic banking services. It is a direct lever for financial inclusion in a country where the informal sector represents nearly 40–50% of GDP.

Responsible AI governance

Deploying these systems requires a rigorous AI governance framework: bias management, explainability of decisions, and compliance with Law 2008-12 on personal data. Data sovereignty and model localization constitute a decisive competitive advantage for actors who invest now.

PI-SPI & PaymentsAugust 2026·14 min

Value Chains Above PI-SPI: The Rail Is Free, the Value Is in the Workflow

On 30 September 2025 in Dakar, the BCEAO launched PI-SPI, the Interoperable Instant Payment System Platform — free for individuals and merchants, instant, and interoperable across all eight UEMOA states. The strategic consequence follows in one sentence: the rail commoditizes the movement of money. The value migrates to what sits above it.

The rail is infrastructure, not a product

PI-SPI is free at the point of use, settles a payment in under 10 seconds, runs 24/7, and connects banks, e-money issuers, microfinance institutions, and payment institutions. Its native primitives — instant push, request-to-pay, scheduled payment, standardized QR, and a centralized alias directory — are a sovereign commodity. There is no per-transaction toll to collect: the transfer rent is removed by decree.

Value migrates to four layers

Four layers stack above the rail. Last-mile UX (USSD, WhatsApp, Wolof) for the informal economy; workflow (invoicing, reconciliation, payroll, collections, subscriptions, escrow, treasury); data turned into credit (scoring, cash-flow forecasting, fraud detection); and Wolof/French AI applied across the three layers below. Each carries rising defensibility — and none can be monetized by a free public utility.

What to build

The wedge is the SME financial OS: invoicing → request-to-pay → auto-reconciliation → cash-flow dashboard. It is the "invoicing + digitalisation of value chains" use-case the BCEAO itself names. Around it: automated collections, mass payroll, marketplace escrow, G2P disbursement, eight-country unified treasury, and WhatsApp commerce where trade actually happens.

What not to build

Four precise traps: a consumer wallet (the account lives at the PSP; no defense), a P2P transfer app (free and already contested), a payment gateway (connectivity is a commodity owned by GIM and the banks), and the rail itself (sovereign, free, built in-house). The test: if your product's only value is moving money, it is already dead.

The Senegal twist

Three factors concentrate the opportunity. USSD + WhatsApp + Wolof remain moats (low smartphone penetration makes last-mile UX a defensible asset). Informality is the opportunity, not the obstacle: 96.4% of employment and ~41.6% of GDP (ANSD); every formalized transaction creates new data that feeds credit. And eight countries, one currency, one integration: cross-border is a single product, not a maze.

The sequencing

Start with the SME financial OS + collections, which generate structured transactional data. Then monetize that exhaust: scoring, BNPL, micro-loans, cash-flow forecasting. Then deploy Wolof-first AI as the compounding differentiator no international firm can match. One caveat: the native API surface (bulk payout, escrow, recurring mandates) is not yet public — the sandbox returns a 404. Whatever is missing from the spec is precisely the buildable surface.

An architecture project to discuss?

Book a Diagnostic