01 / 04§
Engineers as PMs for Claude.
Days of context-gathering, hours of execution. We spend the planning before we let the agent cook — files mapped, constraints written, success criteria explicit.
Vol. 03 · Built for what is next · 2026Harare · Nairobi · Dar es Salaam · KigaliSovereign · Agent-native
Africa's AI-native consultancy
Quantitativerigour.Agenticfuture.Africa.
We architect sovereign AI infrastructure, production-grade agent systems, and high-resolution disease intelligence for African ministries, regulated institutions and global health partners.
Agentic, not autocomplete. Regulator-aware. Built to ship.
Continental Intelligence Surface
Live · Bayesian · Agent-refreshed
Scroll
Trusted across 12+ countries —
Tier-1 BankNational Health AuthorityRegional Economic BodyEnergy SectorResearch CouncilInternational OrgTelecomsPublic Sector
§ 02 — The agentic manifestoWe forget the code.
We forget the code.
Not the product.
The compiler analogy holds. Early users read the assembly. That doesn't scale. At some point the stack absorbs the layer below it — and the work moves up.
Tasks AI can complete are doubling roughly every seven months. The job is no longer typing — it is architecture, product thinking, and continuous orchestration. We build for institutions that understood that already.
02 / 04§
Automate the leaves. Harden the trunk.
Leaf-node features get the exponential. Core architecture, security surfaces and data contracts get human review. Tech debt stays where it can be contained.
03 / 04§
Verifiable by design.
Every change ships with inputs and outputs an auditor can read — eval suites, stress tests and regulator-readable artefacts. Confidence without reading every line.
04 / 04§
Embrace the exponential.
Tasks AI can do are doubling roughly every seven months. Insisting on lockstep review is the new bottleneck. We architect for the model that ships next quarter, not last year.
“Ask not what Claude can do for you — ask what you can do for Claude.”
— Eric @ Anthropic, on agentic engineering■Sovereign AI●Agentic Systems✕Global Health▲Humanitarian◆Logistics & Trade○Conservation▮Agriculture■Banking & Finance●Energy & Mining✕Manufacturing▲Public Sector◆Disease Intelligence○Donor & Multilateral▮Sovereign AI■Agentic Systems●Global Health✕Humanitarian▲Logistics & Trade◆Conservation○Agriculture▮Banking & Finance■Energy & Mining●Manufacturing✕Public Sector▲Disease Intelligence◆Donor & Multilateral○Sovereign AI▮Agentic Systems■Global Health●Humanitarian✕Logistics & Trade▲Conservation◆Agriculture○Banking & Finance▮Energy & Mining■Manufacturing●Public Sector✕Disease Intelligence▲Donor & Multilateral
§ 03 — Practice areasSix pillars.
Six pillars.
One agent-native method.
From sovereign AI infrastructure to disease intelligence and production agent systems — every engagement begins with diagnosis, ends with a system that survives the regulator, the field officer, and the next data refresh.
01 / 06■
Sovereign AI Infrastructure
AI built in Africa, running in Africa, owned by Africa. Air-gapped large-language-model deployments, private RAG, and sovereign agents that never leave your security perimeter.
- —Air-gapped LLM deployment on client-owned hardware (llama.cpp, vLLM, AIBrix)
- —Private RAG and vector search over classified, regulated or proprietary data
- —Data residency by design — national data-protection and sectoral compliance from day one
02 / 06●
Agentic Systems & Automation
Agentic, not autocomplete. We design and ship production-grade agent systems — orchestrated, evaluated, audited — that act on goals across regulated workflows.
- —Production agent design: scoping, tool surfaces, memory, escalation policies
- —Multi-agent orchestration patterns (planner / worker / verifier DAGs)
- —Skills + CLI architectures over MCP-maximalism — debuggable, version-controlled
03 / 06✕
AI-Native Engineering & Data Science
Engineers as PMs for Claude. We architect, plan and ship at the new pace — automated leaf-node delivery, hardened trunks, and changes verified by inputs and outputs an auditor can read.
- —Claude-Code-driven delivery: plan → execute → verify pipelines with eval-first development
- —Predictive ML, NLP and computer vision pipelines deployed at scale
- —Continuous orchestration: architecture, product thinking, multi-agent supervision
04 / 06▲
Data & AI Strategy Advisory
Strategy for the agentic decade. Maturity assessments, sovereign-AI roadmaps, and implementation plans that survive the regulator and the field officer.
- —Data, AI and agent maturity assessments for regulated institutions
- —Tool-agnostic 5-level scoring across strategy, data, models, agents, governance
- —Responsible AI and sovereign-AI roadmaps tied to compliance posture
05 / 06◆
Decision Platforms, Industrial Automation & Bespoke Software
Full-stack decision platforms, industrial automation and bespoke applications — for governments, regulators, operators and industries. From boardroom dashboards to plant-floor SCADA, built to survive the next data refresh.
- —Industrial automation and SCADA / OT integration (water, energy, mining, manufacturing)
- —Bespoke enterprise applications and value-chain automation systems
- —Decision-support dashboards with embedded Bayesian and ML inference
06 / 06○
Disease Intelligence & Geospatial Modeling
High-resolution disease intelligence systems — Bayesian geostatistics, integrated surveillance, and agent-driven pipelines for ministries, research institutes, and global health partners.
- —Bayesian geostatistical risk surfaces for malaria, dengue, NCDs and emerging pathogens
- —Integrated surveillance pipelines (DHIS2, KoboToolbox, EHR, remote sensing)
- —Outbreak detection, R-effective tracking and forecast-to-decision workflows
§ 04 — Sovereign AI for AfricaAI Built in Africa.
AI Built in Africa.
Running in Africa.
Owned by Africa.
Cloud AI is aligned with the company that owns it — not with you. We build sovereign AI that runs entirely within your security perimeter, complies with African data regulations by design, and puts you in complete control of your models, your data, and your agents.
Air-gapped. Data-residency by design.
Local llama.cpp · vLLM · AIBrix.
DSPy · skills + CLI · no MCP-maximalism.
01 / 06◆
Air-Gapped Deployment
Fully isolated on-premises AI. Your data never leaves your infrastructure — by design, not by policy.
02 / 06◆
Data Sovereignty
Built for African regulatory realities — national data-protection laws and sector-specific compliance from day one.
03 / 06◆
Local Model Serving
Production-grade LLM inference on client hardware. Low latency, no API costs, no third-party exposure.
04 / 06◆
Sovereign Agents
Agent systems with full audit trails and human-in-the-loop — operating entirely within your security perimeter.
05 / 06◆
Private RAG
Vector search and retrieval-augmented generation on classified, sensitive or proprietary data. Fully private.
06 / 06◆
MLOps Without Lock-in
Eval, drift detection, lineage and CI/CD on infrastructure you own and control. No cloud captive.
0 bytes
Sent to third-party clouds
100%
Infrastructure ownership
55 AU
Member states in coverage
Day 1
Compliance with national data-protection laws
Ready to own your AI? We deliver a free architecture review, security posture and a 90-day sovereign-AI roadmap.
See the brief→Anthropic
Opus
4.7
4.7
Anthropic Hackathon · April 2026
Qualified · Apr 2026
§ 05 — Built with Opus 4.7We didn't just talk about the agentic future.
We didn't just talk about the agentic future.
We shipped into it.
We qualified into Anthropic's Built-with-Opus-4.7 hackathon under the "Build for what is next" track — shipping agentic research-reproduction infrastructure for African epidemiology in seven days. The same patterns now drive client work.
Claude CodeSub-agentsSkills + CLIEval harnessAfrican epi corpus
Track — Build for what is nextRead the play →
§ 06 — Proof in the metrics
Numbers from the field.
01
0
AU Member States in modelling coverage
02
0 bytes
Data sent to third-party clouds (sovereign deployments)
03
0+
Years of combined founder experience across Africa
04
24/7
Agent-refreshed pipelines — wake at 3am, stay running
“A model is not a model until it survives the regulator,
the field officer, and the next data refresh.”
§ 07 — Selected work
View archive →Systems we have shipped.
Humanitarian × Public SectorDecision Platforms
Regional Risk Intelligence & Anticipatory Action Platform
→Multi-country risk and anticipatory-action platform for disaster preparedness. Fuses hazard, exposure, vulnerability and sectoral data into a single analytical workspace, with AI-assisted reporting and scenario tools — the kind of decision support that lets coordinators reach early-action triggers before the event lands.
Logistics & TradeDecision Platforms
Sovereign Port Operations Platform
→Operations platform for a regulated port system — orchestrating containers, cargo, warehousing and billing across multiple terminals. The national tariff schedule encoded as executable domain logic, real-time SLA tracking, and operational KPIs auto-derived from live data. Enterprise logistics, built sovereign and open.
ConservationDecision Platforms
Satellite Anomaly Detection for Protected Landscapes
→Earth-observation pipeline guarding national parks, game reserves and wetland systems. Satellite indicators are pulled on a regular cadence, compared against a multi-year baseline and surfaced as graded alerts to field teams, authorities and donors — reporting lag compressed from weeks to hours.
§ 08 — The founders
Built by two practitioners.
F01
Co-Founder & CEO
Quantitative epidemiologist and data scientist with a background spanning actuarial science, biostatistics, and public health research across multiple African countries.
Currently pursuing a PhD at the Nelson Mandela African Institution of Science and Technology, developing high-resolution disease risk mapping frameworks using Bayesian geostatistical modeling.
Over a decade driving data analytics, M&E systems and evidence-based strategy — and an active builder in the agentic era, shipping Claude Code workflows, sovereign-AI deployments, and agent-native research infrastructure for African institutions.
Bayesian ModelingDisease IntelligenceAgent OrchestrationR · Python · Stan · gretaSovereign AI
“A model is not a model until it survives the regulator, the field officer, and the next agent-driven data refresh.”
F02
Co-Founder & CTO
Electronic engineer and systems integration specialist with over 12 years designing and deploying industrial control systems, SCADA platforms, and automation solutions across Africa and beyond.
BEng Honours in Electronic Engineering. Deep expertise in PLC programming, process instrumentation, and large-scale infrastructure across water, energy, manufacturing — the systems that fail loudly when they fail.
Registered Professional Engineer in Tanzania and Zimbabwe, with project delivery across Tanzania, Mozambique, Zimbabwe, Qatar and Canada. Now wiring agentic automation into operational technology where uptime matters.
Systems EngineeringSCADA & AutomationIndustrial IoTSovereign InfrastructurePython · SQL
“The systems that matter are the ones that wake up at 3am and stay running — agentic or not. Everything else is a slide.”
§ 09 — BeginBuild the next decade
Build the next decade
of African intelligence
with us.
Disease intelligence. Sovereign AI. Agentic systems. One email reaches both founders. We respond within two working days.