AfriStack Smart System Ltd

Building the Intelligence and Infrastructure Behind the Next Generation of Technology.

AfriStack is an African-founded technology company building AI systems, intelligent software, digital infrastructure, and platforms designed for Africa and the global market.

From proprietary AI development with Omnis to AI-powered products like AfriKDP, Skilluxo and AfriAssist, we build technology that helps people create, learn, work, publish, automate and grow.

Built in Africa. Designed for the world. Powered by intelligence.

We build for
Creators Authors Freelancers Businesses Developers Enterprises
What we build

One technology company. Multiple intelligent systems.

AfriStack develops technology across several interconnected areas — and each one feeds the others.

Artificial Intelligence

We research, train, adapt, evaluate and deploy AI models and intelligent systems designed for real-world applications.

AI Infrastructure

We are developing the infrastructure required to run, integrate, orchestrate and eventually train our own AI models.

Digital Platforms

We build scalable platforms that combine AI, automation, commerce, marketplaces and specialised workflows.

Developer & Business Infrastructure

Our technology is designed not only for our own products, but to provide APIs, AI services and automation capabilities to other businesses and developers.

Our AI ecosystem

The AfriStack intelligence layer.

Our long-term strategy is an interconnected AI ecosystem where our own intelligence powers multiple products, and eventually serves external developers and businesses.

Omnis AI

Omnis is AfriStack's long-term AI development programme: model training, adaptation, AI agents, inference, orchestration, evaluation and intelligent infrastructure.

We have already trained and tested our own models — not as a demo, but as a working pipeline we own end to end: the data, the tokenizer, the architecture, the training loop and the evaluation harness.

omnis-tiny-0.1 · 42M parameters

Omnis Tiny

A language model trained from scratch to validate our end-to-end model development pipeline. Held-out perplexity fell from 237.5 to 41.2 over 15,000 steps and 983M tokens — on a single consumer laptop GPU.

omnis-base · adapter series 1.1 → 1.5

Omnis Base

An adapted open-weight model developed for practical product workloads and experimentation, trained through five successive LoRA adapter generations against our own product benchmarks.

Our objective is to progressively move from experimenting with existing models toward building increasingly capable proprietary intelligence.

Live · afrikdp.com
AfriKDP

AI-powered digital publishing and commerce.

AfriKDP is our digital publishing ecosystem, helping authors and creators publish, sell, distribute and manage digital products — with authors keeping 85% royalties and getting paid across 18+ African countries.

It is evolving well beyond traditional publishing. With AfriKDP AI and Copilot, authors and readers get intelligent tools designed around the complete publishing and reading experience.

AI capabilities

AfriKDP Copilot AI book assistance Book analysis Reading assistance AI tutoring Summaries Quizzes AI narration Writing assistance Translation Intelligent content workflows

The goal is to make publishing and consuming knowledge significantly more intelligent.

Where AI enters the workflow

  • Discover opportunities
  • Create professional profiles
  • Match skills with jobs
  • Create and improve proposals
  • Understand requirements
  • Deliver work
  • Learn new skills
  • Automate repetitive tasks
Live · skilluxo.com
Skilluxo

The intelligent freelancing and services marketplace.

Skilluxo connects professionals, freelancers, businesses and clients through a global services marketplace, with every order protected by secure escrow.

Skilluxo AI is our plan to make AI an intelligent layer across the entire professional services journey — not a bolt-on feature in one corner of the product.

Live · afriassist.com
AfriAssist

AI for creators, businesses and digital work.

AfriAssist brings multiple AI-powered capabilities into one ecosystem, designed around the real workflows of creators and businesses rather than a single AI use case.

AI Assistance

Intelligent conversations and task assistance across the working day.

Content Creation

Generate and transform content for different channels and use cases.

Image Generation

Create visual assets and creative materials without leaving the workflow.

Video Creation

Generate and develop video content and scenes from a written brief.

Music & Audio

Support creative production with AI-powered audio workflows.

Automation

Connect AI with business processes and repetitive tasks.

Creative Workflows

Move from idea → content → production in one environment. AfriAssist plans, drafts and designs autonomously, but publishing stays under explicit human approval unless the user opts into auto-publish.

How everything connects

Our products are not separate experiments.

AfriStack's long-term advantage comes from building an interconnected technology ecosystem, where every product feeds the intelligence layer underneath it.

The AfriStack ecosystem AfriStack splits into AI research and development, which produces Omnis AI, and infrastructure. Omnis AI and the infrastructure layer together power AfriKDP, Skilluxo, AfriAssist and external APIs. Those products generate real-world usage, which produces data and feedback, which in turn improves the AI systems. AFRISTACK AI / R&D research · training · evaluation Infrastructure compute · storage · orchestration OMNIS AI models · gateway · agents AfriKDP publishing Skilluxo marketplace AfriAssist creation · automation APIs external developers Real-world usage how people actually work Data + feedback → better AI systems

Every product creates opportunities to understand real-world workflows, evaluate AI performance, and improve the technology powering our ecosystem.

Our AI development approach

We are not just integrating AI. We are learning to build it.

Our AI journey began with a practical question: can a small technology company build and train its own AI systems with limited resources?

We started with consumer hardware and built our model-development pipeline ourselves — no borrowed training stack, no pretrained weights in the from-scratch line, no external inference API in the loop.

We have run real experiments across every stage of that pipeline:

Data processing Dataset preparation Tokenization Transformer architectures Pretraining Fine-tuning LoRA adapters Model evaluation Product-specific benchmarks AI inference Model orchestration AI deployment

Omnis Tiny 0.1 — measured results

15,000 steps · 983M tokens · ~12 hours on one consumer laptop GPU

MetricStep 500Final
Held-out perplexity237.541.2
Validation loss5.47003.7179
Next-token top-119.9%34.5%
Next-token top-536.9%55.9%
Median fact rank2097

Our early experiments have already produced measurable results. But we view these models as the beginning, not the destination. See the full results →

What we are building toward

From AI applications to AI infrastructure.

Our roadmap extends beyond using third-party AI models.

Today

Integrate and adapt

Integrate and adapt the best available models behind our own gateway.

Next

Specialised models

Develop specialised models and AI systems for our own products.

Then

Proprietary scale

Build larger proprietary models, with specialised African and global language capabilities.

Long term

AI infrastructure

Provide models, APIs, agents and intelligence services to others.

Who that infrastructure serves

AfriStack products Businesses Developers Strategic partners Enterprises Public users

Our ambition is to create technology that can operate independently at increasing levels of the AI stack — and the architectural decision that makes it possible is already in place.

No AfriStack product calls a model directly. Every product calls the Omnis Gateway, and the gateway decides which model answers. Swapping a backend is a configuration change, not a rewrite — so Omnis can take over one capability at a time, as soon as it is good enough.

How the gateway works →
Why AfriStack

Built around real products.

We are not developing AI in isolation. Our AI is tested against real applications across publishing, freelancing, content creation, automation and business workflows.

Build Deploy Test Measure Learn Improve

Inside AfriKDP, our admin tooling measures requests, failures and fallbacks per AI provider — so we can see exactly when Omnis becomes good enough to take a feature back.

Global vision

African foundation. Global ambition.

AfriStack was founded in Africa because we believe exceptional technology can originate anywhere. Our market is not limited to Africa — we build for users, businesses, developers, creators and organisations around the world.

As we enter our next stage of growth, AfriStack is preparing for an international corporate structure and expansion strategy designed to support:

  • Global investment
  • International partnerships
  • Payment infrastructure
  • Enterprise relationships
  • U.S. market expansion
  • Global talent
  • International customers

Where we are strongest

We do not intend to compete with frontier labs on general English reasoning. We intend to be unmatched where they are structurally weak.

Yorùbá, Igbo, Hausa, Pidgin Local context and curricula Cost per token In-region latency Data sovereignty Offline and low bandwidth

Africa is where we started. The global market is where we are going.

Investment

We are building the next phase of AfriStack.

AfriStack is preparing for its next stage of growth and strategic investment. Capital accelerates six things.

AI Research & Development

Larger models, specialised intelligence, training and evaluation.

Compute Infrastructure

GPUs, servers, storage, networking and AI infrastructure.

Engineering

AI researchers, software engineers, infrastructure engineers and product teams.

Data

High-quality datasets, multilingual data and proprietary task data.

Product Development

AfriKDP, Skilluxo, AfriAssist and the broader AfriStack ecosystem.

Global Expansion

International markets, partnerships, enterprise customers and distribution.

Physical Infrastructure

Our long-term vision includes dedicated AfriStack technology and AI infrastructure capable of supporting increasingly sophisticated model development.

Newsletter

Follow what we ship.

We publish a regular briefing on what AfriStack has built, what our AI experiments actually measured, and what is coming next — written for investors, partners and technical readers.

Read the archive →

Get it in your inbox

No marketing filler. Product releases, model results and roadmap changes.

Partner with AfriStack

The future we are building is bigger than one product.

AfriStack is building an interconnected technology company where AI, infrastructure, software, marketplaces, publishing, automation and intelligent applications work together. We have already started. Now we are building the infrastructure to go much further.

Investors Strategic Partners Technology Companies AI Researchers Engineers Enterprises Developers