Services
Workflow automation
If work stops when someone is on leave, or a task only moves because a person remembers to move it, the process is running on memory rather than a system.
Workflow automation replaces the human relay race — the "I sent it to Thandi, she must still approve it" step — with a defined flow that triggers itself, routes to the right person, and records what happened. The work still gets done by people; the difference is that the system carries it between them.
Most South African businesses do not have a process problem in the sense of not knowing what to do. They have a handoff problem: the same information is copied between email, WhatsApp, a spreadsheet and the accounting system, and every copy is a chance for it to be lost, delayed or typed wrong. Automation removes the copies.
Where it does not fit: if the process is genuinely different every single time, or the volume is a handful of cases a month, automation can cost more than it saves. We will say that rather than sell a build. The work worth automating has repetition, a clear definition of done, and someone who currently acts as the transport layer between steps.
Signs this is worth fixing
- Work sits in someone’s inbox waiting for them to remember it
- The same data is typed into two or three systems
- Approvals happen over WhatsApp and leave no record
- Nobody can say where a job currently is without asking a person
- Month-end involves chasing people for the same information
- Processes break when one key person is away
What we automate
Intake and routing
A form, email or new record triggers the flow, which classifies it and routes it to the right queue with the right fields filled in. No one triages by hand.
Approvals with a paper trail
Requests go to the correct approver based on amount, department or type, escalate when they are ignored, and record who approved what, when, and from where.
Document generation
Quotes, invoices, job cards, contracts and handover documents generated from the data already captured, consistently branded and filed to the right place.
Notifications and escalations
The system nudges the person who needs to act, and escalates to a manager when a deadline passes, rather than relying on someone to notice.
Exception handling
The cases that do not fit the pattern get flagged with context, so a person handles the exception instead of manually checking every normal case.
How it works
- 01
Map the process as it actually runs
We sit with the people doing the work and document the real path, including the workarounds nobody wrote down. The map is signed off before anything is built.
- 02
Agree the rules and the edge cases
Who approves what, at which threshold, what happens when the approver is away, and what should stop the flow. These decisions are cheap now and expensive later.
- 03
Build and test with real cases
We build against your live tools, then run a batch of real historical cases through it so failures show up before your team relies on it.
- 04
Launch, train, and tune
We train the people who use it, watch the first weeks of real traffic, and adjust the rules where reality differs from the map.
What you get
- A documented process map, yours to keep
- The working automation, connected to your existing tools
- Approval rules and escalation thresholds written down
- A runbook explaining what to do when something fails
- Training for the people who use it daily
- 30 days of fixes after launch
Systems we work with
We connect whatever you already run — email, WhatsApp, Google Workspace or Microsoft 365, Xero, Sage, QuickBooks, HubSpot, Salesforce, Zoho, Shopify, and most systems with an API. If a tool has no API, we discuss a practical fallback rather than pretending it can be automated cleanly.
Questions people ask
Will this replace my staff?
In our experience it changes what they spend time on. The repetitive capturing and chasing goes away; the judgement, relationships and exceptions remain. Most clients redeploy the hours into work they could not previously get to.
What happens when the automation breaks?
Every build includes error handling: failed runs are logged, the person responsible is notified, and the original data is preserved so nothing is lost. We hand over a runbook and stay available for fixes in the first 30 days.
Do we need to change our tools?
Usually not. We build around what your team already uses, because forcing a new system onto a working team is the most common reason automation projects fail.
How long does a first workflow take?
A single well-scoped workflow is typically live in three to six weeks, including mapping and testing. Larger processes are split into phases so you see value before the whole thing is finished.