Contact us

ResearchIntelligence that works where people are.

We study how to make AI useful and trustworthy in the conditions most research ignores: many languages, modest devices, networks that come and go and systems that must stay secure. And we'll publish what we learn.

Focus
Edge · Language · Security
Partners
OpenAI4Africa · Kalon Edge
Approach
Open by default

Current research

What we're working on now.

Two frontiers: what devices can share when every network is gone, and who answers when AI acts.

In progressPaper forthcoming

Zero bars, still connected?

We're exploring how devices could share data and communicate with no internet, no Wi-Fi, no hotspot and no Bluetooth, using edge technology alone.

With OpenAI4Africa and Kalon Edge

Read the overview

Ongoing

When AI acts, who answers?

Frontier AI is getting better at defending systems and at attacking them. Why we build for accountability, and how every sector can make security part of its AI from day one.

SISU AI Research

Read the overview

Research areas

Where we focus.

Each area connects to the platform, to Alkebulan or to one of our products, so what we learn reaches people.

  • 01

    Edge intelligence

    On-device AI that works offline, on modest hardware and on the power people actually have.

  • 02

    Language models

    Alkebulan, our Pan-African LLM, is multilingual by design. Quadrantica+ makes multilingual NLP efficient, for cost and quality at scale.

  • 03

    AI for cybersecurity

    Frontier AI now defends and attacks. We study how to keep AI systems secure and every AI action accountable.

  • 04

    Responsible AI

    Guardrails tuned with community reviewers to flag cultural misreadings, plus provenance and jurisdiction-aware controls.

  • 05

    Physical AI

    With Pibotix, we study machines that perceive, reason and act in the real world.

  • 06

    Learning and mentorship

    With Altera, we study the safety, inner workings and societal impact of accessible education and intelligent mentorship.

Open by default

We publish what we learn.

Findings, code and resources are shared openly unless there is a clear reason not to. Our six companies each keep a public repository on GitHub.

The AGI approach

Academia. Government. Industry.

Research reaches people faster when the institutions around them are part of it. We work across all three, with partners such as OpenAI4Africa.

Publications and writing

Papers, notes and essays.

Our first paper is in preparation. Until then, here is our thinking, in public.

  1. Forthcoming

    Paper on our offline edge-communication research

    SISU AI with OpenAI4Africa and Kalon Edge · Paper forthcoming

    Overview
  2. October 6, 2026

    Zero bars, still connected?

    Research overview · Edge intelligence

    Read
  3. October 6, 2026

    When AI acts, who answers?

    Research overview · AI for cybersecurity

    Read
  4. October 6, 2026

    If it can't speak Twi, it isn't general

    SISU AI Blog · Research

    Read
  5. October 6, 2026

    Small models, big continent

    SISU AI Blog · Engineering

    Read

Research with us.

Researchers, institutions and organizations: if these questions matter to you, let's work on them together.