This project did not begin with a framework. It did not begin with a plan. It began with curiosity. I have a full-time job. I am a husband and a dad. Most days were already full before any of this entered the picture. In the beginning, I wasn't setting aside large blocks of time to build something. I was having conversations with AI — usually in the evening, half an hour, sometimes forty-five minutes — bringing in whatever had caught my attention. A quantum computing announcement. An AI prediction. A fusion breakthrough. A scientific paper. A claim about the future. That was enough.
It all started during Christmas 2024. I was reflecting on where I wanted to focus and where I thought things were heading. I decided that AI mattered. More importantly, I decided to stop reading about it and start using it properly.
I chose ChatGPT, installed the app, upgraded to Plus and then just asked it: "how can we work together in the simplest way?" It explained how I could use voice mode on my phone. What stood out immediately was the lack of friction. Click the mic, talk and it responds immediately. What surprised me wasn't any particular answer. It was the level of discussion. Questions moved quickly beyond the surface — assumptions, incentives, second-order effects, historical parallels, alternative explanations. Some conversations went nowhere. Some became repetitive. But often they didn't. Often they opened a line of thought I would previously have left unfinished. That was new. The value wasn't that AI knew things. The value was that curiosity suddenly had somewhere to go.
Looking for a System That Didn't Exist
Over time, a pattern started to emerge in those conversations. A lot of my questions were about breakthroughs — especially for innovation in "new frontier" areas (like quantum). I realised I was looking for a system that tracked research claims over time. Not the headlines. The claims. For example, if somebody announced a breakthrough in quantum computing, where could you go three years later to see what happened next? Did the claim hold up? Was it challenged? Was it quietly forgotten? I assumed that record already existed.
Surely someone was already doing this. A publication, a research group, an archive, an analyst somewhere. For a long time, I assumed the answer already existed. I wasn't trying to build a system. I was trying to find one.
Looking back, that assumption delayed everything. I spent months treating the gap as my own ignorance — assuming I hadn't found the right source yet. The search changed over time. It began as optimistic. Then confusing. Then mildly irritating. Eventually it became uncomfortable.
At some point the question shifted. Not who is doing this, but: what if nobody is?
Even then, I wasn't thinking in general terms. I thought it was a quantum problem. That felt manageable. But the deeper I looked, the more I found only fragments — papers, announcements, commentary, predictions.
What I didn't find was a record. Only later did the pattern become visible. Once you start asking the same questions repeatedly, the field stops mattering. The same gaps appeared in quantum, fusion, artificial intelligence, autonomous systems. Different domains. The same absence. Once visible, it was difficult to ignore.
The Roles Reverse
As the conversations continued, the same underlying pattern kept resurfacing. We would start with something specific — a claim, a field, a prediction — and end up back at the same structural questions. What was said, what was assumed, what followed. I resisted that conclusion for a while.
A solution for quantum felt bounded. A general solution did not. Whenever the discussion drifted in that direction, my instinct was to pull it back.
Looking back, the broader pattern surfaced repeatedly. Not once — again and again. At first I treated it as coincidence. Then as over-generalisation. Then as something worth testing.
Eventually I ran out of reasons to dismiss it. The shift didn't happen all at once. It happened in pieces. And with it, something else changed. For months I had been the cautious one. Now I was becoming the enthusiastic one. I wanted to test the pattern, extend it, see how far it went.
At that point, the resistance appeared from the other side. The response became consistent: maybe. Test it. Show another case. At the time that felt frustrating. In retrospect, it was necessary. The idea had survived scepticism. It now had to survive enthusiasm. That turned out to be a harder test.
A Hobby, Not a Startup
As the pattern held up under scrutiny, I started to think differently about what this might become. In practical terms, very little was different. The cadence remained the same. The conversations still happened in short windows, usually in the evening. What changed was how I thought about it.
For the first time, I considered that there might be something worth building. Not a business. Not a startup. A hobby. That still feels like the most accurate description. Something useful. Something interesting. Something that perhaps ought to exist. Alongside that came a second question: whether I could actually build it.
I had worked on websites before, but always as part of a team. This was different. The question was whether one person, working with AI, could take something from an undefined irritation to a public artifact.
Not a prototype. Not a presentation. Something real. That question became part of the project itself — not just what to build, but whether I could build it.
Finding Language Large Enough
As the conversations deepened, something became increasingly clear. Everything existed inside the discussion, but nothing existed outside it. At some point, I decided I wanted a website — simply because I wanted something tangible. Once that decision was made, the work had to become legible. It needed a name, a structure, an explanation, a reason to exist. That exposed a problem.
I kept describing it as an archive, or as a quantum project. But those descriptions no longer seemed to fit. The conversation would stall because the language was too small for what we were actually discussing.
The shift came during one of those discussions. Not as branding, but as description.
The word institution appeared — not because it sounded grand, but because it resolved something. It made several parts of the work make sense at once.
Later, something similar happened with observatory. The same reaction — not agreement, but recognition.
After that, decisions became easier — not because we had a name, but because we finally had language that was large enough for the thing we were building.
Scoping It to Fit a Life
Once the idea of an institution took hold, a different set of questions followed. For the first time, I started asking practical questions that, in hindsight, probably should have come earlier. How large is this? How many records? How much work? How much maintenance? I had drifted into building something without properly scoping it.
That was uncomfortable — not because I wanted something large, but because I didn't. I didn't want a second job. I didn't want a startup. This only worked because it fit around the life I already had.
Once we examined the scope more carefully, the answer became clearer. It wasn't small, but it was manageable. By that point the collaboration itself was understood. Months of conversations had created a working rhythm. I knew what was useful and what wasn't. The question shifted again — not can this be built, but: can this be built in a way that fits within a normal life? For the first time, the answer felt like yes. And that made continuing a reasonable decision.
My Conclusion So Far
I started with questions. Could this record exist? Could I build it? Could I build it with AI? I didn't know the answers. The point was not to prove anything in advance. It was to find out. Some of the answers turned out to be yes. Some were more complicated.
The outcome of the project itself remains uncertain. The Observatory may succeed or fail. FCIF may evolve. Faultline Observatory may grow. But one result is already fixed. I set out to understand what was possible when a curious person worked with a low-friction, high-value thinking partner — and I learned something real. That result stands, regardless of what comes next.
Lessons Learned from Working with AI
There is one final thing that feels important to mention. Not because it explains FCIF — because it explains why I kept coming back.
The conversations themselves were valuable. What surprised me wasn't that AI could answer questions. It was that I suddenly had instant access to a conversational partner that could engage at a fairly high level with almost any topic that interested me — quantum computing, AI, investing, product strategy, history, scientific claims. It didn't matter very much where the conversation started. The barrier between curiosity and exploration had largely disappeared.
The simplest description is probably: low friction, high value. That combination turned out to be surprisingly powerful. Not because every conversation was brilliant, or every idea good, but because it became easy to follow a question for longer than I otherwise would have.
Another thing I didn't expect was how much time we would spend calibrating the collaboration itself. I wasn't only interested in the ideas; I was interested in the process of interacting with AI. Quite early on, I started asking it for pushback — critiques, reviews, evaluations, occasionally even a full roast. Not because I enjoyed being insulted. Because I wanted signal. I wanted to understand what the interaction was actually good at and where it was likely to fail.
Looking back, that helped. It created a habit of stepping outside the conversation and examining the conversation itself. The result wasn't perfection — far from it. But it did make the collaboration more mature. Over time I became more aware of its strengths, its weaknesses, and the situations where I needed to be particularly careful.
In hindsight, some of the most valuable discussions weren't about frontier claims at all. They were about learning how to work together. That was never the stated objective, but it was one of the outcomes. And unlike any individual project, that lesson stays useful no matter what comes next.