Intelligence
Thinking on AI infrastructure.
Analysis, takeaways, and operational notes for enterprises building AI systems.
Stop Automating the Work That Makes You Money
Agencies keep pointing AI at the copy, the imagery, the video. That's the actual product clients are paying for. The right target is the friction around it: briefing, research, and brand-voice checks.
Read articleWhy Most Agencies Are Stuck in the AI Sandbox
Plenty of agencies have a Slack channel full of AI experiments and a few enthusiasts doing clever things alone. Almost none have turned that into a business-wide capability. Three reasons, and none of them is the tooling.
The Mirror Principle: Agencies Can't Guide a Shift They Haven't Made
Brands are moving from search rankings to LLM citations, from one brand voice to individual personalisation, from broadcast to real-time dialogue. Agencies still running last decade's playbook can't take clients through this.
DoorDash Didn't Buy an AI Platform. It Built One.
DoorDash launched an internal AI research lab and its own benchmark, tuned on its own codebase and incident history. The lesson isn't about DoorDash's scale. It's about the question every business should be asking.
SA Brands Are 18 Months Behind. That Is the Opportunity.
The US job market is about to be reshaped by AI. South Africa has a buffer. Most brands are using it to wait instead of move.
AI Summit London 2026: Key Takeaways
Agentic AI, data quality barriers, AI ROI pressure, and the shift from pilots to platforms. Our read on what the 2026 London AI Summit means for South African enterprises.
South Africa's Internet Is Still Mostly Human. That Window Is Closing.
Cloudflare data shows 65.5% of SA internet traffic comes from humans, against a global pattern where bots now account for 57.3% of all HTTP requests. Here is what it means.
When the Board Asks About Your AI: What You Need to Have Ready
AI governance has moved from IT policy to board agenda. What auditability, transparency, and responsible AI mean in practice, and what a POPIA-compliant AI deployment actually looks like.
Why Your AI Pilot Worked and Your Rollout Will Not
The transition from proof of concept to production AI infrastructure is where most projects stall. The failure modes are predictable. So are the fixes.