AI Employees Platform
Agentic core
The base platform our agents run on. Multi-LLM model garden, native RAG, visual workflow tooling, omnichannel delivery, full audit trail.
9 agents in production todayFIG 1.0 · THE CAPABILITY STACK
Five partners.
Sixteen tools.
One delivery contract.
We hold direct contractual partnerships with five platforms. You inherit the full capability under a single delivery contract — we carry the partner-management complexity, you get the certified tools, configured for South African enterprises.
5
Platform partners
Direct contracts. Quarterly product reviews.
16
Tools we implement
From AI agents to low-code to voice IVR.
1
Delivery contract
Partner complexity stays our side of the table.
100%
Certified engineers
On every engagement, in every stack we ship.
Stubber
stubber.com ↗Python gives you a working agent. An orchestration layer gives you a defensible one. Your CFO does not need to know the difference, until they need to know the difference.
Workflow Studio
Build surface
Visual designer for agent logic and conversation flows. We design the agent here; clients can read the same diagram.
Every agent designed here first
Model Garden
LLM routing
Multi-LLM selection per step — flagship models for reasoning, smaller models for classification. Cost-per-task as a design parameter, not an afterthought.
Cost-per-task baselined at MVP
Channel Connectors
Channels
WhatsApp Business API, Voice, email, web chat. The format your customer already uses — no portal, no app to install.
WhatsApp-first for SA marketXOPS
xops.io ↗
XperiencOps Spoke
IT automation
ServiceNow-certified spoke for autonomous resolution of IT tasks — onboarding, licence lifecycle, configuration, access governance. No RPA. No scripts. Goal-oriented AI.
Cencora: 1.25M actions/yr · 92% autonomy
Copilot Connector
Microsoft integration
Native connector to Microsoft Copilot — surfaces XOPS autonomous actions inside Teams and M365 without leaving the Microsoft environment your people already use.
Corteva: post-carve-out Copilot integration
ISO/IEC 42001 Governance
Compliance
ISO/IEC 42001 (AI), 27001:2022, 27018, SOC 2 aligned. CISO-accepted Configuration-as-Code under US financial services regulation. Short procurement conversation.
S&P Global: CISO-accepted under US finregMatogen
matogen.com ↗
Carys
NL analytics
Natural-language analytics over your data warehouse. Ask the question, get the chart — with row-level security and audit trail attached to every query.
Self-service BI for business teams
Mendix
Low-code
Enterprise low-code platform for the apps your agents need to talk to. Build the surface, ship the back-of-house workflow, deploy under your tenancy.
Bespoke ops apps in weeks
Altair RapidMiner
Predictive analytics
Predictive modelling and ML pipelines for the workflows where a classical model still outperforms an LLM — churn, propensity, risk scoring.
Decisioning models for SA marketMicrosoft
desireinfoweb.com ↗
Copilot Studio
Agent build
Build agents that live inside Teams, Outlook, SharePoint, and your custom Power Platform surfaces. Native to where your people already work.
M365-native agent deployments
Azure AI Foundry
Model platform
Enterprise model orchestration — deploy, evaluate, and govern AI inside Azure under your existing data-residency, identity, and compliance posture.
Production AI inside Azure tenancy
Power Platform
Automation
Power Automate flows, Power Apps surfaces, Dataverse persistence — the glue layer that makes Copilot Studio agents actually do things in your systems.
Operational glue layer
SharePoint Intelligence
Knowledge
Turn the document library you already have into a defensible knowledge base for your Copilot agents — with permission-scoping inherited from SharePoint itself.
RAG over your real document estate
Copilot Connector
Integration
Connect Copilot to the line-of-business systems Microsoft doesn’t cover natively — in-house ERPs, custom CRM, regulatory reporting stacks.
Bridge to non-Microsoft systemsNuacem
nuacem.com ↗
Botjet
Conversational AI
Production conversational AI for customer-facing channels at scale. The platform behind enterprise WhatsApp deployments — carrier-grade throughput and uptime.
Customer-facing AI at carrier scale
Convojet
Voice / IVR
Hybrid IVR — AI on the front line for deflection and classification, human handoff with full context when the conversation needs it.
Contact-centre deflection at scale
Observejet
Quality & analytics
Live conversation monitoring, sentiment, escalation triggers, and post-conversation analytics — the quality layer most agent deployments forget to build.
Continuous quality monitoringFIG 1.1 · THE POSTURE
What we do. What we don’t.
We do
Hold the partnerships.
Five direct contractual relationships. Quarterly product reviews with each. Certified engineers on every engagement. You inherit the capability under one delivery contract.
We do
Configure for South Africa.
POPIA alignment, ZAR billing, SA enterprise integrations, local data residency where the platform supports it, WhatsApp-first channel fit.
We don’t
Build the platforms ourselves.
The platforms above are the work of teams ten, fifty, two-hundred-engineers strong. Re-creating their work would dilute the discipline and the outcome. We deploy them; we don’t reinvent them.
FIG 1.9 · WHICH STACK FITS YOUR PROBLEM
The right
partner per problem.
The first conversation is about your problem, not our platforms. We’ll point you at whichever of the five (or which combination) actually fits — even if “none of the above” is the honest answer.