Edge AI, built in Africa

AI that works where the grid doesn’t.

Load shedding, patchy networks and remote sites shouldn’t put AI out of reach. Syntu runs AI on site, on little power, with or without a network, and keeps your data where it belongs.

Dock 3 · InboundLive
Supplier
Example Foods (Pty) Ltd
Manifest
24 pallets
Counted
22 pallets
Evidence
3 clips · 14:32:07 to 14:41:55
Short2 pallets short. Claim pack ready for the supplier.

Illustration with example data.

The problem

AI is built for places with reliable power and fast networks.

Much of Africa can’t count on either. Load shedding, patchy coverage, remote sites, heat, theft and tight budgets leave three gaps.

Power Needs the grid

Most AI depends on constant electricity and a link to distant data centres. Load shedding switches it off when it is needed most.

Connectivity Needs a network

Cameras and sensors that stream to the cloud fail on remote sites, underground and wherever coverage drops, and the data costs add up.

Control Run elsewhere

Sensitive footage and data travel offshore to be processed, under other countries’ laws and other companies’ terms.

The platform

Built from the edge inward.

We start where AI meets the real world: on site, on little power, with or without a network. Local data, models and infrastructure grow behind it.

Concept: a Syntu data centre with rooftop solar, linked by fibre to a warehouse, a clinic and an office tower, each running AI at the edge. 3 2 1
Concept Syntu at work: AI on site at warehouses, clinics and offices, backed by local infrastructure
  1. 1Edge intelligence: AI running on site, on little power
  2. 2African data and models: governed and kept local
  3. 3Sovereign compute: local infrastructure behind it all
Layer 1

Edge intelligence

AI running on site, on low-power hardware, processing data where it is created. Works through load shedding and with or without a network.

Building now
Layer 2

African data and models

The operational data and adapted models that edge deployments generate, governed and owned locally.

Next
Layer 3

Sovereign compute

Local infrastructure that coordinates, updates and supports thousands of edge devices, built as demand grows.

Later
First product · Warehouse vision AI

Your cameras already see every pallet. Now they can count them.

Syntu turns the cameras at your dock doors into a verification system. Every delivery is counted, checked and timestamped as it happens, and the AI runs on site, so footage never leaves the building.

  • Count pallets in and out and match them against the delivery note or manifest.
  • Spot damage and broken seals at the door, before stock is signed for.
  • Keep timestamped evidence for supplier claims and disputes, ready to share.
  • Work with what you have: existing IP cameras, edge devices on site, and through load shedding.
Why on site

AI that keeps working, and stays yours.

Works off the grid

Edge devices run on little power and keep working through load shedding and patchy connectivity. Models are tested on African operations, not imported assumptions.

Your data stays here

Processing happens on your site, or in South African data centres when it needs to. Personal information stays under POPIA, not someone else’s terms of service.

Open foundations

Open models on open software, on hardware we can change as better chips arrive. No dependency on a single vendor, and no surprise price changes.

Build it with us

Looking for people who want to build AI that works anywhere, starting here.

Syntu is early. If you share the mission, there is room to help shape it from the start.

Builders

Engineers and a technical co-founder

People who have run models in production: edge AI, computer vision, embedded and low-power systems, ML infrastructure. We are looking for a co-founder to lead the technology.

Partners

Researchers, labs and hardware partners

Edge AI and vision teams, universities and research labs, hardware makers, integrators and infrastructure operators who would rather build together than alone.

Backers

Early investors

Angels and impact investors who believe the next wave of AI will run at the edge, and want to back it from the beginning.

Get in touch

Write to the address below and tell us how you would like to be involved.

Run a pilot

See it on your own dock in eight weeks.

We are choosing a small number of retailers, wholesalers and distributors for paid pilots. One site, a handful of dock cameras, and results measured against your own claims and counts.

  • 1 site
  • 4 to 8 dock cameras
  • 8 weeks
  • Agreed metrics
Pilots, partners and people hello@syntu.ai