Transitioning to enterprise software. Services live now. First product, RAG Studio, ships Q4 2026. See the roadmap →
03 Cluster · Intelligence Operations Layer

From raw data
to acted-on decisions —
in one closed loop.

Analytics Studio, Data Pipelines, and Workflow Engine — three products that close the loop between insight and action. Visual ETL with no SQL required for business analysts. Embedded BI for executives. Workflow automation that fires the moment a metric crosses a threshold. POPIA data lineage tracked from the first pipeline stage.

Data analysts Data engineers Operations Finance
Product surfaces
3
Analytics Studio · Data Pipelines · Workflow Engine
Source connectors
200+
Sage · Xero · SAP · Salesforce · PostgreSQL · Snowflake · Kafka
Workflow integrations
500+
Trigger-action automation across the apps your team already uses
Source service lines
2
Service 13 — Data Analytics · Service 15 — Hyper-automation & RPA
Why the Data & Analytics OS exists

Insight without action is a wall poster. Action without insight is theatre.

South African data teams stitch together Snowflake, Fivetran, dbt, Looker, Zapier, and a UiPath licence — paid in dollars, governed by no one, audited by accident. Cluster 03 was built specifically to dissolve those four problems with one platform.

01 — The insight-action gap

Dashboards report what just happened. Nothing happens next.

Most analytics ends at the chart. The dashboard flags a 14% drop in conversion, an exec sees it in the morning, an analyst opens a ticket — and the response sits in a backlog for a week. Data without a trigger is a wall poster, not an operating system.

2 stacks Most teams run analytics and automation as completely separate vendors. The handoff is human.
02 — POPIA without lineage

Where did that field come from? Nobody knows.

When a data subject access request arrives, you have 30 days to show every system that processed their personal information and the legal basis for each transformation. Most data warehouses cannot answer that. They were not designed to.

30 days POPIA Section 23 response window for data subject access requests — non-negotiable.
03 — SQL as a gatekeeper

Every chart needs a data engineer. Every data engineer is busy.

Marketing waits two weeks for an attribution chart. Operations waits a sprint for a stock-out report. The bottleneck is not insight — it is the SQL skill required to extract it. Business analysts who could ship the chart in an hour cannot get past the join.

14 days Typical wait time between a business question and the dashboard that answers it.
04 — Process opacity

You know there are inefficiencies. You can't point at them.

Every operational team has automation candidates buried in their ticketing logs, their email threads, and their forms. Process mining surfaces them — but the established tools are priced for Fortune 500 budgets and never speak ZAR.

3–5× Typical efficiency gain when a manual process is mined, scored for ROI, then automated.
Vendor stitched stack

The old way

  • Fivetran + dbt + Looker + Zapier + UiPath — all in USD
  • Lineage tracked nowhere — POPIA requests answered manually
  • Insight in one tool, action in another, gap between them
  • Business analysts blocked behind SQL gatekeepers
  • Data residency drifting wherever each vendor stores it
sonofgraig

The unified OS

  • One subscription in ZAR — predictable, in-budget
  • POPIA lineage written by every transformation, by default
  • Threshold-triggered workflows close the insight-action loop
  • No-SQL drag-and-drop dashboards for business analysts
  • All queries land in af-south-1 — proof, not promise
Five-layer architecture

A data warehouse, a BI tool, and an orchestrator — wired into one operating model.

The Data & Analytics OS sits between Cluster 01's AI and Cluster 02's infrastructure. It feeds the AI products with clean, governed data, and exposes the cloud spend, security findings, and pipeline health from Cluster 02 as first-class analytics surfaces.

L1
Consumers
Every department, not just engineering. Finance, operations, marketing, HR, and the executive team — each with role-aware views.
Data analysts Operations Finance Marketing Execs
L2
Products
Three user-facing surfaces drawn from two service lines. Independently subscribable, but the loop only closes when all three run together.
Analytics Studio Data Pipelines Workflow Engine
L3
Data plane & lineage
Shared lakehouse, shared catalog, shared lineage graph. Every transformation writes its source, purpose, legal basis, and destination.
Data Catalog Lineage Manager Row-level security
L4
Engines & orchestration
Apache Airflow for scheduled DAGs, dbt for transformations, Kafka for streaming, Metabase + Superset for embedded BI, n8n core for workflow execution.
Apache Airflow dbt Kafka Metabase n8n core
L5
Storage & compute
PostgreSQL primary, Snowflake or Redshift for warehouse workloads, Iceberg-format lakehouse for raw layers — all hosted in af-south-1 by default.
af-south-1 PostgreSQL Iceberg 99.9% SLA
The closed loop — why this cluster compounds

Analytics tells you what is happening and why. Automation acts on that insight without requiring human intervention for each decision. Together they form a closed loop: data surfaces a pattern, automation responds to it. Sold as separate products, the gap between them is bridged manually. Sold as one OS, the loop closes automatically — and the longer it runs, the more processes get folded in.

Three product surfaces

Analytics. Pipelines. Workflows.
One OS, one lineage graph.

Each product solves a specific bottleneck for a specific persona — but they were designed to compose. Analytics Studio reads from Data Pipelines. Workflow Engine triggers from Analytics Studio thresholds. Document Processing feeds the AI agents in Cluster 01.

Product 01 · Service 13 — Data Analytics

Analytics Studio

Embedded BI for operational and executive reporting. No SQL required for standard dashboards.

Q2 2027 Subscription + per-seat embed
Drag-and-drop dashboards
Build operational dashboards without SQL. Pivot, filter, and chart with direct manipulation of fields.
50+ live data connectors
Sage · Xero · SAP · Salesforce · HubSpot · Google Analytics · Postgres · Snowflake — all live, all rand-priced.
AI chart recommendations
Powered by Cluster 01 — the studio reads your dataset shape and proposes the chart type that best surfaces the pattern.
Row-level security
Users see only their permitted records. Department-level, region-level, customer-level — defined once, enforced everywhere.
Embedded analytics
Expose dashboards to your customers under your brand. Available on Growth and Enterprise tiers, priced per embedded seat.
Acquisition & attribution
Pre-built marketing analytics module. Multi-touch attribution, channel ROI, and cost-per-conversion — across paid and organic.
User journeys & retention
PostHog integration for product analytics. Funnel analysis, retention cohorts, feature flag impact tracking.
Threshold alerts
Set a threshold on any chart. The Workflow Engine fires when it crosses — sending Slack, email, or any of 500+ downstream actions.
Business analysts
Drag-and-drop dashboards in minutes, not sprints.
CFOs & finance
Self-serve P&L, cash, AR, and consolidations.
Marketing
Multi-touch attribution and channel ROI in rand.
Product teams
PostHog-powered funnels and retention cohorts.
Product 02 · Service 13 — Data Engineering

Data Pipeline Builder

Visual ETL/ELT design with POPIA lineage written automatically by every transformation.

Q2 2027 Tier + data volume metering
Visual ETL designer
Drag a source. Drag a destination. Connect them with transformation nodes — joins, filters, aggregations, model calls.
200+ source connectors
Databases, SaaS APIs, file systems, and streaming sources. Sage · Xero · SAP · Salesforce · Postgres · Kafka · S3.
POPIA lineage
Every transformation records source, purpose, legal basis, retention, and destination. Lineage maps export as PDF or JSON.
Scheduled & event-triggered
Run on cron, on file arrival, on webhook, on Kafka topic, or chained off another pipeline's success.
dbt-as-a-service
SQL transformations in the dbt convention with version control, testing, and documentation generated automatically.
Streaming ingestion
Kafka and Kinesis ingestion at sub-second latency for click-stream, IoT, and operational telemetry workloads.
Data quality tests
Inline assertions: not-null, unique, accepted values, referential integrity. Failed tests block downstream consumers.
Data Catalog
Searchable inventory of every dataset, with column-level descriptions, owners, and PII classification.
Data engineers
Visual canvas with dbt and Airflow under the hood.
Information officers
POPIA lineage at the click of a button. DSAR-ready.
Data architects
Catalog, lineage, and quality in one shared workspace.
Analytics engineers
SQL transformations with tests and version control.
Product 03 · Service 15 — Hyper-automation & RPA

Workflow Engine

500+ app integrations with visual workflow design — every execution logged with POPIA processing basis.

Q3 2027 Tier + run metering + doc volume
Visual workflow builder
Trigger-action automation with branching, looping, conditional logic, and human-in-the-loop approval gates.
500+ app connectors
Slack, Teams, Gmail, Outlook, Salesforce, HubSpot, Stripe, Xero, Sage — and the long tail your team depends on.
Trigger taxonomy
Schedule, webhook, file drop, email arrival, Kafka topic, dashboard threshold — the loop closes wherever your data lives.
Document Processing
OCR + classification for invoices, contracts, KYC packs, claim forms. Structured output flows into your ERP or warehouse.
RPA Bot Studio
Record-and-replay automation for legacy systems without APIs. Bridges mainframe greenscreens, Citrix sessions, and old web apps.
Conversation Routing
Route inbound conversations from email, web chat, WhatsApp, and SMS to the right Cluster 01 agent or human queue.
Human-in-the-loop
Define approval gates on consequential actions. Route approval requests via Slack, email, or in-app inbox.
Process Discovery
Mines system logs to surface automation candidates ranked by ROI — calculated before a single workflow is built.
Operations leads
Automate the repetitive 80% — keep humans on the 20%.
Finance & AP
Invoice OCR + 3-way match + journal posting in one flow.
HR teams
Onboarding workflows: ticket, IT provisioning, payroll, sign-off.
Customer ops
Inbound routing, ticket triage, and AI agent escalation.
What it plugs into

200 source connectors. 500 workflow apps.
The data and the action are already where you keep them.

The Data & Analytics OS does not require you to migrate. It reads from the systems your business already runs in — Sage, Xero, SAP, Salesforce, Postgres — and writes back into the apps your team already uses every day.

Finance & ERP connectors

South African finance teams live in Sage, Xero, and Pastel — and increasingly the larger ones live in SAP. The OS reads transactional data, GL accounts, and AR/AP positions live, with no nightly export gymnastics.

Sage One · Sage 200 Evolution · Sage X3
Xero · QuickBooks Online · FreshBooks
Pastel Partner · Pastel Xpress
SAP S/4HANA · SAP Business One · SAP ECC
Oracle NetSuite · Microsoft Dynamics 365 F&O
SARS eFiling exports · VAT201 · IRP5/IT3

CRM & marketing connectors

Pre-built attribution and pipeline analytics across the channels your marketing and sales teams already pay for. Multi-touch attribution, channel ROI, and lead-to-cash flow — out of the box.

Salesforce · HubSpot · Pipedrive · Zoho CRM
Google Ads · Meta Ads · LinkedIn Ads · TikTok
Google Analytics 4 · Mixpanel · Amplitude · PostHog
Mailchimp · Brevo · Klaviyo · Customer.io
Shopify · WooCommerce · Magento · Takealot Marketplace
Stripe · PayFast · Yoco · Peach Payments

Databases & streaming sources

Point Data Pipelines at any database, warehouse, lake, or stream. CDC (change data capture) is supported on Postgres and MySQL. Streaming sources land at sub-second latency.

PostgreSQL · MySQL · MariaDB · MSSQL · Oracle
MongoDB · DynamoDB · Cassandra · Redis
Snowflake · BigQuery · Redshift · Databricks
Apache Kafka · AWS Kinesis · GCP Pub/Sub · Azure Event Hubs
S3 · GCS · Azure Blob · SFTP · FTP
CDC streams via Debezium · log-based replication

Operations, comms, and the long tail

Workflow Engine reaches the apps your team already lives in. The library covers the obvious (Slack, Teams, Gmail) and the necessary (WhatsApp Business, signing tools, ticketing systems, HRIS).

Slack · Microsoft Teams · Discord · Mattermost
Gmail · Outlook · Zoom · Google Meet
WhatsApp Business · SMS gateways · Twilio · Vonage
Jira · Asana · Monday · ClickUp · Trello · Linear
DocuSign · SignNow · Adobe Sign · PandaDoc
BambooHR · Sage People · Workable · Greenhouse
POPIA lineage by default

Three audit questions every data team gets eventually.
The OS answers all three at runtime.

When a data subject access request lands, you have 30 days to respond. POPIA Sections 23 and 24 require you to identify every system that processed personal information, the legal basis, and the recipients. The Data & Analytics OS records this on every transformation — not retroactively.

Where did this field come from?
Every column in the warehouse can be traced back through every transformation, every join, every filter — to the raw source row that produced it. With timestamps and the user who built the pipeline.
→ Data Pipelines · Lineage graph
What was the legal basis?
POPIA Section 11 grounds — consent, contract, legal obligation, legitimate interest — recorded as metadata on every transformation. The basis travels with the data through the entire pipeline.
→ Data Pipelines · POPIA basis tagging
Who received the data after?
Every dashboard view, every workflow execution, every embedded analytics seat is logged with the user, organisation, and purpose. DSAR responses are generated, not assembled.
→ Analytics Studio · Access audit
Where the closed loop earns its keep

From insight to action — six examples from real South African operating models.

Cluster 03 is the broadest cluster by buyer base — every department has analytics needs and automation candidates. The patterns below recur across legal, financial services, retail, healthcare administration, and manufacturing.

Financial services
Anomaly detection on a fraud trigger flow
Analytics Studio threshold flags a 5σ spike in declined transactions. Workflow Engine fires: alert SOC, freeze the merchant ID, open a Jira ticket, send an SMS to the relationship manager — all within 30 seconds.
Dashboard threshold Workflow Engine SARB-aligned
Retail & e-commerce
Stock-out automation across Takealot and Shopify
Pipelines pull live inventory from Shopify, Takealot, and the warehouse WMS into one fact table. The Studio surfaces stock-outs by SKU. Workflow Engine triggers reorder POs and alerts the buyer when reorder points are crossed.
Data Pipelines Multi-source merge PO automation
Professional services
Time-and-billing recovery dashboards
Live feed from time-tracking, AR, and project budgets into a partner-level realisation dashboard. Each partner sees their own data only (row-level security). When realisation drops below 70%, Workflow fires a coaching prompt.
Row-level security Embedded Coaching loop
Healthcare admin
Medical aid claim throughput with POPIA lineage
Document Processing reads claim forms via OCR, extracts structured fields, and posts them to the practice management system. Every record carries POPIA lineage tagged with Section 26 (special personal information) handling.
Document AI POPIA s.26 Lineage
Manufacturing
OEE telemetry with predictive maintenance triggers
Kafka ingests PLC telemetry from the floor at sub-second latency. Studio dashboards show OEE in real time. When vibration anomalies are detected, Workflow triggers a maintenance ticket and reschedules the production plan.
Streaming ingest Predictive trigger OEE
HR & people ops
Onboarding and offboarding orchestration
Hire event in BambooHR or Sage People triggers a workflow that provisions IT, requests Jamf MDM enrolment, generates a payroll record in Sage, posts a welcome message to Slack, and books the orientation calendar slot.
Multi-app workflow Sage + Jamf + Slack HRIS triggers
Honest roadmap

When each Data & Analytics product ships.

Cluster 03 is a Phase 2 build. Analytics Studio and Data Pipelines launch alongside Cluster 02's Cloud Console. Workflow Engine and Document Processing follow in Q3 2027. Process Discovery rounds out the cluster in Q4.

Phase 1 — Months 1–6 — Funded by AI Dev Platform
Architecture & data plane design
In progress
Lakehouse + warehouse architecture
POPIA lineage schema (OpenLineage spec)
Data Catalog metadata model
Connector framework (Singer + Airbyte CDK)
Visual ETL canvas prototype
Embedded BI runtime selection (Metabase + Superset)
Funded by: Phase 1 platform revenue
Progress: Architecture in design
Phase 2 — Months 7–9 — Cluster entry
Analytics Studio + Data Pipelines
Q2 2027
Drag-and-drop dashboard builder — no SQL required
50+ live data connectors (Sage, Xero, SAP, Salesforce)
Visual ETL designer with 200+ source connectors
POPIA lineage on every transformation
Row-level security and data catalog GA
dbt-as-a-service for analytics engineers
Target: First Data & Analytics customer live
Cross-sell: AI Dev Platform + Cloud Suite customers
Phase 2 — Months 10–12 — Cluster expansion
Workflow Engine + Document Processing
Q3 2027
Visual workflow builder with branching and looping
500+ app connectors GA (Slack, Sage, Salesforce, etc.)
Threshold-triggered workflows from Analytics Studio
Document Processing — invoice, contract, KYC OCR
Conversation Routing into Cluster 01 Agent Builder
Embedded analytics for Growth + Enterprise customers
Streaming ingest GA (Kafka, Kinesis, Pub/Sub)
Target: 50 Cluster 03 customers
Certifications: SOC 2 Type II + POPIA
Embedded analytics ARR: R5M+
Phase 2/3 — Months 13–15 — Final product
Process Discovery + RPA Bot Studio
Q4 2027
Process mining from system logs — automation candidates ranked by ROI
RPA Bot Studio — record-and-replay for legacy systems
Mainframe greenscreen, Citrix, and old web app bridges
Marketing analytics module (multi-touch attribution)
Product analytics module (PostHog integration)
Marketplace for community-built connectors and templates
Dependency: Workflow Engine GA
Cross-sell: UiPath / Celonis displacement
Pricing for the cluster

One subscription. Priced in rand. Plus volume.

Data & Analytics OS is included in Growth and Enterprise tiers. Volume metering applies on data scanned (Analytics Studio), pipeline rows processed (Data Pipelines), and document pages (Workflow Engine). Embedded analytics is per external seat.

Cluster 03 — Data & Analytics OS

Bundled with Growth, included in full at Enterprise.

Growth includes any three clusters — pick Data & Analytics with AI Dev Platform and Cloud & DevOps for the complete operations OS. Enterprise includes all five clusters plus Governance Hub. Standalone Data & Analytics tier opens at the entry of Phase 2.

R19,999
Growth tier · 25 seats · billed monthly

Insight, action,
and the loop between them.

Cluster 03 ships from Q2 2027 onwards, funded by Phase 1 platform revenue. Joining the early interest list secures founding-customer pricing and direct input into the cluster's connector roadmap.

POPIA lineage on every transformation 200+ source connectors 500+ workflow integrations af-south-1 by default