appsHuman-Centred AI

AI That Works
On Your Terms.

We build enterprise applications where you stay in command. AI models handle the reasoning — your data stays yours, your rules and infrastructure remain enforced, your outcomes stay predictable.

appsExplore Our Apps schemaHow It Works buildBuild Your App
construction
Custom Built App
psychology
AI Harnessing
security
Security Layer
manage_accounts
Human Control
100%Data Ownership
ZeroVendor Lock-in
Self-HostedLocal · Shared · Dedicated
POPIACompliant Architecture
appsOur Applications

AI-Powered SaaS,
Owned by You.

Six production applications, each purpose-built to harness AI within a human-controlled, security-first architecture. All self-hosted on your own infrastructure.

arrow_forward
PaymentFlow
Payment Intelligence
AI-assisted billing, invoicing, expense management and financial reporting. Automates the tedious, surfaces the anomalies, and keeps your finance team in control of every approval decision.
AI Reconciliation Audit Logs Human Approval
arrow_forward
ContentFlow
Content Operations
AI content calendar planner and approval system. Schedule posts, generate blog articles from any content using Claude, GPT-4o, Gemini, or Grok. Share secured live client feeds with login.
Brand Guardrails Editorial Review 6 Networks
arrow_forward
VoxGen
AI Workstation
Multi-provider AI workspace powered by Claude, OpenAI, Gemini, xAI and loads of open-source locally hosted models. Chat, document intelligence, image studio, content studio, and knowledge bases.
AI Generation Open Source Models Local & Cloud
arrow_forward
VoxCloud
Unified Cloud File Manager
Connects Google Drive, OneDrive, FTP/SFTP, and your locally hosted filesystem into one slick, secure interface. Full file manager, cross-provider transfers, public sharing, and in-browser editor.
Google Drive OneDrive FTP / SFTP
arrow_forward
VoxSecure
Security Operations
OWASP codebase scanning meets real-time threat monitoring. 52 scan rules, 21 threat detectors, A01–A10 full OWASP coverage. One self-hosted dashboard. Zero external dependencies.
OWASP Scanner Threat Monitoring Compliance Ready
arrow_forward
VoxIntel
Enterprise Intelligence
Connect every application database — finance, security, cloud, AI, marketing — and surface live unified intelligence in one beautiful engine. Visual widget builder, AI reports, role-based access.
Multi-DB Connect AI Reports Live Dashboards
Build Your AI App
Custom Development
Have a workflow that needs AI intelligence? We design and build bespoke applications with the same human-in-the-loop, security-first architecture that powers all our products — on your infrastructure, with your data, under your control.
rocket_launchStart the Conversation read_moreOur Process
schemaOur Process

The Human-in-the-Loop
Pipeline.

Every AI interaction in our applications follows a structured flow — human intent drives the process, AI amplifies capability, and security gates ensure nothing escapes your control.

01 — INTENT
cognition
Human Intent

You Define the Goal

Every process begins with a human operator setting clear intent — what to do, what boundaries to respect, and what acceptable output looks like. No AI acts autonomously.

personRole-based access tuneConfigurable scope
02 — SANITISE
privacy_tip
Input Guardrail

Sanitise & Scope

Before anything reaches an AI model, our application layer strips sensitive data, enforces prompt boundaries, and validates the request falls within your configured policy. PII stays out.

privacy_tipPII stripping policyPolicy validation
03 — REASON
smart_toy
AI Reasoning

Model Does the Heavy Lifting

The scoped, sanitised prompt is dispatched to the AI model. It reasons, structures, and generates output — but only within the lanes you've defined. Context, tone, format all controlled by you.

apiAny LLM backend data_objectStructured output
04 — VALIDATE
fact_check
Output Guardrail

Validate Before It Lands

Raw model output passes through a validation layer — schema checks, content policy enforcement, and format verification. Hallucinations get flagged, not silently passed through.

schemaSchema checks warningAnomaly flags
05 — DELIVER
task_alt
Human Delivery

You Own the Outcome

Validated output is surfaced to the operator through your application's UI. Review, approve, edit, or reject — the human stays in the loop at every meaningful decision point.

rate_reviewReview & approve storageStored on your server
verifiedCore Philosophy

Built on Six
Non-Negotiables.

These aren't aspirational values — they're architectural decisions baked into every line of code we ship.

manage_accounts
Human Override Always
No automated AI action is irreversible without human sign-off. Your people stay in command of every meaningful decision. AI is the copilot, not the captain.
storage
Your Data, Your Server
We don't aggregate your business data. It lives on infrastructure you control. We build the application layer — the data pipeline belongs to you, always.
model_training
Model-Agnostic Architecture
Our applications abstract the AI layer. Swap between Claude, GPT, Gemini, or a local model without rewriting your workflows. No lock-in to a single AI provider.
security
Security by Default
Every application ships with POPIA-aware data handling, role-based access control, full audit logging, and prompt injection defences baked into the stack from day one.
visibility
Full Transparency
Every AI call is logged with the prompt sent, model used, tokens consumed, and output received. Nothing in our applications is a black box — you can audit everything.
build
Custom-Built, Not Off-the-Shelf
We don't reskin generic AI tools. Every application is architected for your specific workflow, industry requirements, and operational constraints. Tailored fit, not one-size-fits-all.