The AI Layer · Khanyisa Computer Systems

The POS that sees what your business is doing.

Cloud-first point-of-sale with AI sales intelligence — built for restaurants, canteens, campuses, and multi-site operators across South Africa. Ten categories of intelligence that legacy POS vendors don't have, backed by real-time dashboards, exportable reports, role-based views, and multi-site consolidation.

Live

Real-time dashboards across every site and till

Export

Reports for finance, operations, and owners

Intelligence categories (not a reports menu)

≥99.5%

POS uptime, written into every SLA

The product

One platform. Three products. Every touchpoint.

The AI module is the upsell that turns a Standard subscriber into a Professional subscriber. It sits on top of the reporting foundation: live dashboards, exportable reports, role-based views, and multi-site consolidation.

Core Platform

The operational backbone

POS integration, customer & account balances, stock + orders, operational reports. The system of record for the business.

Premium AI Module

The intelligence layer

Forecasting, anomaly alerts, branch comparisons, menu engineering, boardroom narratives — turns the raw POS stream into decisions managers can act on the same day.

Enterprise Services

The margin driver

Multi-site rollouts, custom integrations, SLA-backed support, managed hosting. What turns a software sale into a long-term relationship.

The moat

Ten categories of intelligence that legacy POS vendors don't have.

Every AI feature on the platform maps to one of these ten buckets. Nothing invented outside this list.

Executive Intelligence

Revenue velocity, run-rate forecasts, same-store growth, margin alerts.

Variance Detection

Unusual declines, silent failures, hidden opportunity bundles.

Product Intelligence

Momentum score, cannibalisation, menu engineering matrix.

Time Intelligence

Peak slots, staffing recommendations, dead-hour promotions.

Customer Intelligence

Top spenders, dormant customers, repeat-visit cadence.

Operational & Fraud

No-transaction alerts, void patterns, revenue leakage.

AI Forecasting

Next week / next month sales, demand prediction, payday effects.

Branch League Tables

Rank by growth, basket, profit. Diagnose underperformers.

Boardroom Narratives

Auto-generated management summaries with actions and risks.

Control & Finance

Cash vs card ratios, duplicates, reconciliation exceptions.

The maturity framework

What's live. What's shipping. What's coming.

Three labels, applied next to every AI feature mention site-wide. ROADMAP features never appear on the homepage hero — only here and on product pages.

LIVEToday

Working in production. Book a demo and see it.

AI Reporting AssistantPlain-English narratives, KPI snapshot, sentiment-tagged insight cards. Rule-based today; example below is real output.
Menu Engineering matrixEvery item classified Star / Plough Horse / Puzzle / Dog from live sales data.
AI item-image generationPhoto-realistic menu imagery via OpenAI, called only on user click. Item name & description in — no sales data.
Plain-English report queriesNatural date ranges, POS codes, cost centres. No SQL, no BI analyst.

IN BUILDQ2–Q3 2026

Engineering is actively shipping these.

Demand forecastingCategory-level volume per hour per site, 7 days ahead. Production planning, staffing, stock ordering become data-driven.
Operational anomaly detectionVoids, discount usage, cash-up variance flagged in near real-time. Catches fraud and training gaps before losses.
Price elasticity modellingRecommended pricing per item per site. Pricing decisions backed by evidence, rolled out centrally.
Voice-driven POS"Two flat whites, one with oat milk" → order captured. Drive-thru, prep stations, accessibility.

ROADMAP12–18 months

Planned, designed, directionally committed.

Autonomous menu optimisationMenu engineering + price elasticity + demand forecasting feed an optimiser that suggests (and with permission, executes) menu changes continuously.
Supply chain intelligenceIngredient-level cost tracking, automatic substitute flags, recipe-level margin recalculation as input costs move.
Customer segmentation & personalisationPersonalised menus, offers, and pricing for CuisineOnline accounts based on order history.
Multi-lingual voice and orderingEnglish, isiZulu, Afrikaans, Sesotho, Portuguese, French, Swahili — pan-African deployment.

The AI Reporting Assistant

Not demo-ware. The real report a site manager gets in their inbox.

Below is an actual rendered AI report from a Khanyisa site (Site ID redacted). Same shape, same sentiment tags, same suggested actions. This is what "the POS that sees" means in practice.

AI Reporting Assistant Report

Site ID: ▮▮ · Period: 2025-08-01 → 2027-01-31 · Comparison: 2024-01-30 → 2025-07-31

Large window mode: monthly trend view

Manager summary

Across 8 month(s), strongest month was 2025-12 (R 77,682.70 gross); softest was 2026-03 (R 2,811.40). Gross moved -96.3% across the window.
Month-over-month gross changes that crossed ±10%: 2025-11 → 2025-12: +14.2%; 2025-12 → 2026-01: -12.7%; 2026-02 → 2026-03: -95.8%.
Recorded POS gross sales total R 503,531.32 across 13,464 transactions (avg ticket R 37.40). That is -58.6% versus the prior period (R 1,215,632.44).
Strongest day was 2025-09-25 (R 4,868.10); softest day was 2026-02-08 (R 22.00).
Total POS sales for the window moved -58.6% vs the prior period.
Strongest POS trade band is lunch (12:00–14:00) at about R 201,007.70 gross.
Lunch window gross changed by -59.1% vs the prior period.
Sales from 15:00 onward underperformed by about 59% vs the prior period.

POS gross sales

R 503,531.32

POS transactions

13,464

Avg ticket (POS)

R 37.40

Monthly sales trend

info

Across 8 month(s), strongest month was 2025-12 (R 77,682.70 gross); softest was 2026-03 (R 2,811.40). Gross moved -96.3% from the first month in range to the last.

Month-over-month signals

negative

Notable threshold (±10%) crossings: 2025-11 → 2025-12: +14.2% (notable); 2025-12 → 2026-01: -12.7% (notable); 2026-02 → 2026-03: -95.8% (major).

Suggested: Review promotions, seasonality, and operations for months with large swings.

Daily sales summary

warning

Over the selected period, recorded POS gross sales total R 503,531.32 across 13,464 transactions (avg ticket R 37.40). That is -58.6% versus the prior period (R 1,215,632.44).

Sales spread by day

info

Strongest day was 2025-09-25 (R 4,868.10); softest day was 2026-02-08 (R 22.00).

Period trend

warning

Total POS sales for the window moved -58.6% vs the prior period.

Suggested: Review staffing, promotions, and product availability for the weaker period.

Peak service window

info

Strongest POS trade band in this range appears to be lunch (12:00–14:00) at about R 201,007.70 gross.

Lunch trade

warning

Lunch window (12:00–14:00) gross changed by -59.1% vs the prior period.

Suggested: Review lunch menu, throughput, and staffing if the drop persists.

Late-day trade

warning

Sales from 15:00 onward underperformed by about 59% vs the prior period.

Significant sales decline

negative

POS gross for the period is down 58.6% compared with the prior period — worth a manager review.

Suggested: Validate data completeness, then review trading hours, pricing, and operations for the period.

Online orders

info

No online staging orders in this window, or summary procedure returned no rows.

Generated: 2026/04/20, 13:55:34

The philosophy

AI that makes operators smarter, not redundant.

Boring beats flashy.

A plain-English daily summary beats a chatbot nobody uses.

Rules first, models second.

External LLMs only when deterministic rules can't do the job. Cheaper, faster, auditable.

Your data stays yours.

Inference happens on infrastructure you (or we) control. No third-party training on your sales data.

Commercial model

Three tiers. Recurring by design.

The AI module is the wedge that drives Standard → Professional. Boardroom narratives and branch league tables drive Professional → Enterprise.

Standard

R 2,500/ month

Single-site operators. Core POS + reports + basic online ordering.

POS integration
Daily sales reports
Customer accounts
Email support

How it works under the hood.

Hospital and university procurement asks the same questions. Here are the answers, plainly.

Today: rule-based analytics.

Your data is queried on your infrastructure. We do not send raw transaction data to external services.

For AI image generation:

OpenAI's API, called only when a user clicks "Generate image". Inputs are an item name and description — never sales or customer data.

For upcoming model-driven features:

Evaluating private-hosted options (Azure OpenAI, AWS Bedrock) so customer data stays within your regulatory boundary.

POPIA / GDPR:

Full data residency and processing details on the legal page. No vendor lock-in — your data is yours, always exportable.

Written into every SLA

Local support in SA business hours
POS uptime ≥ 99.5%, measured monthly
Response under 2 hours for critical faults
Monthly AI insights pack delivered to your inbox
No vendor lock-in — your data is always exportable

Ready when you are

30-minute demo. No slides. Just the assistant running on your numbers.

We'll review your current POS setup and show the AI insights running against your real data. No obligation.