Build, deploy, fine-tune, and govern AI from one POPIA-native environment. Four product surfaces — Agent Builder, RAG Studio, Fine-Tuning Ops, and Governance Hub — sharing one API gateway, one model registry, and one observability stack. Hosted in AWS af-south-1. Priced in rand.
Three structural problems make AI difficult to operationalise inside regulated African enterprise. The AI Dev Platform was built specifically to dissolve them.
Today's AI stack means model APIs from one vendor, agent orchestration from another, vector storage somewhere else, monitoring duct-taped on top, fine-tuning compute from a fifth provider, and compliance done manually in spreadsheets.
Banks, insurers, government bodies, and healthcare cannot deploy AI they cannot explain or audit. Most AI tooling has no native governance layer — so compliance departments block deployment indefinitely.
AWS, Azure, and Google Cloud have infrastructure in af-south-1 — but no POPIA documentation, no local compliance support, no ZAR pricing, and no understanding of South African enterprise procurement cycles.
Enterprise AI teams routinely waste two-thirds of their capacity on infrastructure plumbing, integration glue, and compliance rework — not on the intelligence layer that creates business value.
The AI Dev Platform launches as a focused product, matures into a platform, and opens as an ecosystem. The same sequence followed by Salesforce, Shopify, and Microsoft Azure. Five layers separate concerns and create the data flywheel that makes the platform compound over time.
Every interaction generates signal: prompt performance data, evaluation results, agent execution traces, bias detection results, and usage telemetry. That data — stored with full consent and governance — feeds back into model quality, default templates, evaluation benchmarks, and recommendations. The longer a customer uses the platform, the better the platform performs for them. This is the same loop that made Google Search unassailable and Microsoft 365 difficult to replace.
Each product is independently subscribable. Each solves a specific problem for a specific persona. Together, they share an API gateway, a model registry, and an observability stack — making the platform more than the sum of its parts.
A visual environment for designing, testing, and deploying autonomous AI agents.
Private knowledge chat systems grounded in your enterprise data — with citations.
Train and manage domain-specialised AI models on your proprietary data — without it leaving af-south-1.
Audit, monitor, and govern AI systems for enterprise compliance — POPIA, GDPR, EU AI Act, ISO 42001.
Beneath the four product surfaces sits a shared platform services layer. This infrastructure is built once and serves all products simultaneously — what makes the platform more than the sum of its parts.
One authentication surface across all four products. Single login. Single API key. Single audit trail. RBAC, rate limiting, and tenant isolation built in.
Central catalogue of foundation models and customer fine-tunes. Performance benchmarks, bias evaluations, A/B routing, and controlled deprecation — all in one place.
Token usage, latency, errors, prompts, drift, and cost — tracked across every product. Hooks into Datadog, Grafana, and PagerDuty for teams with existing observability infrastructure.
The most strategically important component. The data layer is what separates a feature from a platform — and a platform from a defensible business. Everything happens here, with consent, governance, and tenant isolation enforced at the kernel.
South African banks, insurers, government, and healthcare cannot deploy AI they cannot explain or audit. Compliance teams block deployment indefinitely without these three answers. Governance Hub is the unlock.
We publish this because enterprise buyers deserve to know the actual state of the product they are evaluating. Services revenue funds the build. We hit our dates.
Starter includes one cluster — pick AI Dev Platform and get RAG Studio. Growth includes three. Enterprise includes all five plus Governance Hub. No USD pricing. No FX risk.
Get RAG Studio, 5 seats, 50 GB document storage, 500K LLM tokens per month, audit log, and full POPIA compliance documentation. Ideal for an enterprise proof of concept with real data.
3 founding customer slots remaining. 247 organisations on the waitlist. Apply for early access — we onboard the first cohort in person across Pretoria, Johannesburg, and Cape Town.