Intelligent infrastructure

Axioms

The intelligent infrastructure arm of Indices Technology — conversational AI, machine intelligence, edge computing, IoT and secure digital infrastructure, built as one shared layer instead of isolated AI systems bolted onto each product.

Talk to Dasah See the architecture

Dasah · capability trace

one interface, many systems

Asked

"Where is my order from the Kaduna seller, and is that seller verified?"

Resolved

  • marketx.get_order
  • marketx.get_seller
  • logistics.track_delivery

Answered

Order #4821 left the Kaduna hub this morning and is out for delivery. The seller is verified, with a trust score of 92.

What Axioms is

Rather than building isolated AI systems for each product, Axioms provides a shared intelligence layer through which applications expose capabilities, exchange events and access intelligent services.

At its centre is Dasah — a universal conversational interface through which people and systems interact with capabilities across the Indices ecosystem. MarketX is one of the first clients of this infrastructure; future Indices platforms connect to the same underlying intelligence.

How it fits together

Architecture

Dasah is not a chatbot for one product. It is the conversational layer of Axioms, and every Indices platform is a client of the same infrastructure. 

Axioms
   │
   └── Dasah  — conversational layer
        │
        ├── MarketX
        ├── Logistics
        ├── Security
        ├── IoT
        ├── Edge systems
        └── Future Indices platforms
Axioms — the infrastructure
The shared intelligence and interaction layer of Indices. It owns the capability registry, the event fabric, identity and the intelligent services every platform draws on.
Dasah — the conversational layer
The universal interface to Axioms. People and systems speak to it in text, voice and eventually other modalities; Dasah works out what is needed and reaches for the right capability.
Clients — the platforms
MarketX, logistics, security, IoT and edge systems connect as clients. MarketX is the first major use case for Dasah, not its owner — new platforms plug in without rebuilding it.

The contract

Capabilities, not tightly coupled apps

Every system publishes what it can do as a named, versioned capability. Dasah never needs to understand the internals of MarketX, a logistics hub or an edge device — it asks Axioms for the capability. New Indices products can be added without rebuilding Dasah.

marketx.*
  • get_order
  • get_seller
  • get_trust_score
logistics.*
  • track_delivery
  • get_route
  • assign_dispatch
security.*
  • get_alerts
  • get_incident
  • verify_access
iot.*
  • get_sensor
  • list_devices
  • set_threshold
edge.*
  • get_device_status
  • deploy_model
  • sync_state
knowledge.*
  • search_documents
  • summarize
  • cite_source

The nervous system

Event-driven by design

Systems in the ecosystem emit events rather than calling each other directly. Those events trigger workflows, notifications, analytics and AI reasoning — which is what turns a set of applications into an ecosystem instead of a web of hidden dependencies.

  • order.completed
  • seller.verified
  • payment.completed
  • delivery.updated
  • device.alert
  • trust.score.changed

A deliberate separation

Intelligence is not orchestration

An automation platform is not the brain. Dasah decides what should happen; a workflow layer makes sure it happens reliably.

Decides

Dasah

Understands intent, holds conversational state, selects capabilities and reasons over results.

  • Intent understanding across text and voice
  • Conversation and session state
  • Capability selection and tool calling
  • Reasoning over results and context
  • Access decisions on behalf of the user

Executes

Workflow layer

Durable execution, retries, scheduling and long-running processes — replaceable, and never the foundation of the intelligence itself.

  • Durable, retryable execution
  • Long-running and scheduled processes
  • Fan-out to downstream services
  • Visual automation for external integrations
  • Swappable — the intelligence does not depend on it

What we build

Capability domains

The disciplines Axioms covers — deployed inside the Indices ecosystem, and available to partners who need the same infrastructure.

Conversational AI

The universal interface layer — text and voice interaction that resolves intent and routes it to capabilities across the ecosystem.

  • Text and voice
  • African languages
  • Tool calling
  • Session state

Machine Learning & Forecasting

Models that predict equipment failure, demand surges and yield outcomes before they happen — trained on operational data from across the group.

  • Anomaly detection
  • Demand forecasting
  • Maintenance scheduling

RAG & Knowledge Systems

Retrieval-augmented reasoning over private documents, SOPs and databases, so answers carry citations instead of confidence alone.

  • Vector retrieval
  • Source citation
  • Role-based access

Edge AI

Local model execution on farm, mining and field hardware — decisions that survive a lost connection and sync when the link returns.

  • On-device inference
  • Offline-first sync
  • Remote deployment

IoT & Telemetry

Real-time telemetry from sensor fleets — soil health, livestock vitals, vehicle position, facility conditions — normalised into shared event streams.

  • MQTT device fabric
  • Device registry
  • Threshold rules

Computer Vision

Visual inspection, quality control and surveillance automation running at the edge, where the cameras are.

  • Defect detection
  • Object counting
  • Real-time alerting

Document Intelligence

Extraction, classification and summarisation of invoices, contracts, customs papers and reports — at the volume real operations produce.

  • OCR + LLM extraction
  • Clause analysis
  • ERP integration

Secure Digital Infrastructure

Identity, authorisation, auditability and observability across every capability call — security as a property of the platform, not an afterthought.

  • Auth and access control
  • Audit trails
  • Model and data isolation

APIs, Integration & Automation

Versioned APIs, capability schemas and event contracts that let internal platforms and third-party systems join the ecosystem cleanly.

  • Versioned APIs
  • Capability schemas
  • Workflow automation

Want your systems to speak this language?

Book a scoping call and we'll map your data, capabilities and events.

Talk to us

Built to be inherited

Engineering principles

What we build now is infrastructure that future teams will inherit. The goal isn't a heavyweight enterprise architecture on day one — it's avoiding the hidden coupling that becomes a nightmare later.

  • 01 Clear service boundaries
  • 02 Explicit contracts
  • 03 Versioned APIs
  • 04 Capability & tool schemas
  • 05 Event contracts
  • 06 Authentication & authorisation
  • 07 Observability
  • 08 Auditability
  • 09 Documentation
  • 10 Replaceable components
  • 11 Database ownership boundaries
  • 12 Architecture Decision Records

Technology direction

The stack, and why

Languages are chosen for the work they do, not for fashion. The Axioms backend is a service in its own right — interfaces are clients of it, never its foundation.

TypeScript core
Axioms and Dasah live where APIs, realtime communication, tool calling, auth, events and workflows meet — the work TypeScript is genuinely good at.
Services · Realtime · Tool calling
Interfaces as clients
Nuxt and Vue front-ends are clients of the Axioms backend, not its foundation. The intelligence is reachable from any surface, not trapped in a web app.
Nuxt · Vue · Decoupled
Python for ML
A specialised layer for machine learning, computer vision and scientific workloads — behind clear service boundaries rather than woven through the core.
ML · Vision · Research
Go and Rust at the edge
Introduced where edge, networking, security or performance requirements genuinely justify them — and not before.
Edge · Networking · Performance