Stop Buying AI Tools. Start Building an AI Strategy.
There’s a pattern playing out across South African businesses right now that’s burning money and building frustration in equal measure. It goes like this: leadership reads about AI, gets excited, buys Copilot licences for 200 users, announces the rollout in a town hall, and then waits for the magic to happen.
Three months later, adoption is sitting at 12%. The finance team tried it once and went back to their spreadsheets. Sales says it’s “interesting but not useful.” IT is fielding complaints about Copilot surfacing documents people weren’t supposed to see. The CFO starts asking uncomfortable questions about ROI.
This isn’t an AI failure. It’s a strategy failure. And it’s happening everywhere.
The tool-first trap
Microsoft has made AI remarkably accessible. Copilot is embedded in the apps your people use every day. Azure OpenAI Service lets you build custom AI solutions on your own data. Dynamics 365 is shipping autonomous agents that can process invoices, triage service tickets, and generate demand forecasts without human input.
The technology works. That was never the question. The question is whether your organisation knows what it’s trying to achieve with it.
Most don’t. Not because they’re behind or lacking ambition — but because nobody stopped to ask the basic questions before the licences were purchased. What processes are costing us the most time? Where is human effort being wasted on repetitive work? Which data sources are trustworthy enough to feed an AI model? Who in the organisation is ready to adopt this, and who needs training first?
Without answers to those questions, AI becomes a solution in search of a problem. Expensive, underwhelming, and politically toxic when the board asks what they got for the spend.
What AI Jumpstart actually is
Braintree’s AI Jumpstart programme exists because Chris Badenhorst got tired of watching this pattern repeat. As VP of Azure at Braintree, he’s spent years deploying cloud infrastructure and AI workloads for South African organisations. The consistent finding: the ones that succeed with AI are the ones that did the strategy work first. Every time.
AI Jumpstart is a structured engagement that takes an organisation from “we know we need to do something with AI” to “we have a working proof of concept, a clear roadmap, and a governance framework” in weeks rather than quarters. It’s not a sales pitch disguised as a workshop. It’s a proper diagnostic followed by a build.
Here’s how the phases work:
Phase 1: Strategy Assessment. This is where the uncomfortable conversations happen. What are your actual business objectives? Not “we want to use AI” — that’s not an objective. What are the measurable outcomes you need? Cost reduction? Faster time to insight? Better customer experience? Reduced manual processing in finance? The strategy assessment maps your motivations to realistic AI outcomes and identifies who in the organisation needs to be involved.
Phase 2: AI Readiness Assessment. This is the one most organisations skip and then regret. Braintree’s team evaluates five dimensions: your business strategy alignment, your current AI maturity, your data foundations, your governance and security posture, and your organisational culture. That last one matters more than most people think. An AI deployment into a team that doesn’t trust the technology or understand its limitations will fail regardless of how good the model is.
Phase 3: Planning and Adoption Workshop. Based on the assessment findings, Braintree works with your team to build an adoption plan, evaluate skills gaps, set up governance frameworks and FinOps controls, and identify the first AI workload. This isn’t a theoretical exercise. The output is a concrete plan with timelines, owners, and dependencies.
Phase 4: Architecture and Security Design. Before anything gets deployed, the landing zone goes in. This is where Braintree’s Azure expertise matters. The AI infrastructure is built on Microsoft security best practices from day one — not bolted on after someone raises a concern six months later. Identity, access controls, data classification, network architecture — all designed before the first model runs.
Phase 5: Proof of Concept Deployment. This is where theory becomes reality. Braintree deploys a working proof of concept against a real business use case identified in the earlier phases. Not a demo. Not a sandbox exercise. A functional implementation using your data, your processes, and your people. The POC gives leadership something tangible to evaluate and gives the broader organisation a concrete example of what AI looks like when it’s done properly.
Why skipping the strategy costs more than the programme
The maths on this are straightforward. A Copilot licence runs around R540 per user per month. Deploy that to 200 users without a strategy, and you’re spending over R100,000 a month on a tool that 80% of your people aren’t using effectively. Over a year, that’s R1.3 million in licence costs alone, before you count the IT support time, the change management effort, and the opportunity cost of a failed rollout making the next AI initiative harder to get approved.
AI Jumpstart costs a fraction of that wasted spend. More importantly, it ensures that when you do deploy Copilot — or Azure OpenAI, or Dynamics 365 agents — you’re deploying into an environment that’s ready. Clean data. Clear use cases. Trained users. Proper governance. Measurable outcomes from day one.
The organisations getting real value from AI in South Africa right now aren’t the ones with the biggest budgets. They’re the ones that started with a strategy instead of a shopping list.
The security dimension people forget
There’s another reason the strategy-first approach matters, and it’s one that doesn’t get enough attention: security. Every AI deployment expands your attack surface. Copilot indexes your SharePoint, your OneDrive, your email. Azure OpenAI services connect to your data estate. Dynamics 365 agents interact with your ERP and CRM data.
If your permissions model is a mess — and after years of organic growth, most are — AI will happily surface sensitive data to people who shouldn’t see it. That’s not a hypothetical. It happens in the first week of most unplanned Copilot deployments.
AI Jumpstart’s architecture phase addresses this directly. The security design happens before the AI deployment, not after the first data leak. Working alongside Braintree’s security practice and Arctic Wolf’s 24/7 security operations, the programme ensures your AI infrastructure is monitored, governed, and locked down from the start.
Getting started
If your organisation has been circling around AI without committing — or worse, has committed without a plan — AI Jumpstart is the reset button. It works for organisations at any stage: those that haven’t started, those that started and stalled, and those that deployed something and aren’t seeing results.
Chris Badenhorst and the Azure team at Braintree run these engagements across industries — manufacturing, financial services, retail, professional services, public sector. The programme adapts to your environment, your data maturity, and your ambition.
You don’t have an AI problem. You have a strategy problem. AI Jumpstart solves the strategy so the technology can do what it was built to do.
Contact Braintree at info@braintree.co.za or visit braintree.co.za/ai-jumpstart to start the conversation.
