We've been watching AI for years before it became a boardroom topic. These six posts trace how our thinking evolved in real time, from measured scepticism about GPT-3 to the night everything changed. No manufactured foresight. Just honest reactions as the technology moved faster than anyone expected.
The First Signal
November 2020. GPT-3 had been out for a few months. The demos were impressive. The enterprise applications were not. Mak ran it through our standard evaluation criteria and the conclusion was clear: fascinating technology, premature for regulated industries. The gap between demo and deployment was enormous.
Chapter 1 - Nov 2020. Measured scepticism.
GPT-3 and the Question Nobody's Asking
Take·6 min read
The Pace Quickens
Eighteen months later, AI wasn't standing still. NLP had improved. Computer vision was finding real traction in quality control. Robotics was advancing in logistics. Our annual fringe tech review captured a feeling we couldn't quite name: things were moving faster than we'd expected. Enterprise applications were still limited, but the technology trajectory was steeper than any of us had predicted.
Chapter 2 - May 2022. The acceleration.
Fringe Tech Report: What We're Watching in 2022
Report·6 min read
Open Source Changes the Game
Then Stable Diffusion landed. Open-source image generation. The technical achievement was genuine, and the enterprise use cases were modest. But the real significance was what it meant for AI accessibility. When capability moves from API-gated to open-source, the pace of innovation changes fundamentally.
Chapter 3 - Sep 2022. The open-source shift.
Image Generation Just Changed (Stable Diffusion First Impressions)
Take·5 min read
Something Is Building
A month later, we zoomed out. Integration, cloud maturity, and intelligent automation were reshaping enterprise tech. AI was progressing faster than expected. But nobody could quite articulate what it meant. This post captured that feeling: the sense that something significant was coming, without knowing exactly what.
Chapter 4 - Oct 2022. The build-up.
The Next Five Years of Enterprise Tech
Perspective·7 min read
The Night Everything Changed
December 5th, 2022. Isaac created a ChatGPT account out of professional curiosity. Expected to spend twenty minutes. Spent three hours. The conversational interface was the breakthrough: not a tool you operated but a colleague you talked to. Still wrong about things. Still limited. But the baseline capability was a step change from anything we'd used before.
Chapter 5 - Dec 2022. The inflection point.
I Just Spent Three Hours Talking to a Computer
Perspective·6 min read
What It Actually Means
Ten days later, the initial reaction had settled into something more systematic. We tested ChatGPT against real enterprise work: content drafting, code scaffolding, technical explanation, decision support. Strong for some. Dangerous for others. The conclusion wasn't certainty. It was the most productive uncertainty we'd felt in a decade.
Chapter 6 - Dec 2022. The assessment.
What ChatGPT Means for Enterprise Software
Article·8 min read
Why This Series Exists
Reading these posts in order tells you something you can't get from a retrospective: what it actually felt like to watch AI arrive. The scepticism was real. The surprise was real. The uncertainty was real. We didn't know in November 2020 that we'd bet the company on AI by June 2023. Nobody did.
If you're trying to understand how a technology-sceptical enterprise team ended up going all in on AI, this is the arc. Not the polished version. The honest one.
Reading these posts back, you can see the exact moment my scepticism broke. It wasn't a gradual shift - it was a Tuesday night in December, three hours in, realising I had no idea where this goes.
Isaac Rolfe
Managing Director