The Next Ten Years Are Built Before They Arrive
A reflection on twelve years of annual reviews, the cost of waiting for certainty, and building systems that reduce the gap between seeing and acting.
In April 2026, I fed twelve years of annual reviews into my AI agent system and asked one question:
What pattern have I been repeating without noticing it?
The answer was uncomfortable.
I often see the right direction. Then I wait too long.
In 2017, I entered crypto near the top — early enough to see it, late enough to pay tuition. In 2023, I found SNEK after months of due diligence. The instinct was right. The entry came after enough confirmation to feel safe.
That is the pattern.
Pressure first. Pain second. Breakthrough third. Action after confirmation.
Sounds rational from the inside. Also expensive.
The problem was rarely intelligence. It was a bias toward certainty. In markets, life, and career — certainty usually arrives after the best price has already passed.
The good news: a pattern you can name becomes a pattern you can redesign.
The system behind the delay
Over the past months I built Hermes and Samantha. They read reports, monitor the portfolio, scan signals, write drafts, build tools, and remember decisions I would otherwise forget.
The goal: reduce the gap between seeing and acting.
If I can turn “one step late” into “half a step late,” the next ten years change a lot.
Different tools. Same belief.
A system that does not depend on willpower beats effort that depends on mood.
Six paths, no clean plan
I think in scenarios, not forecasts. Ten years is too long. Markets break assumptions. Family needs shift. AI will make today’s workflows look primitive.
So — six paths, none of them clean:
The steady sovereign individual. Keep building in public. Agent system improves. Portfolio stays balanced: crypto, yield, gold, stocks, enough cash to protect family decisions. Not dramatic. Compounds.
The big cycle. If BTC, SNEK, or the broader crypto cycle delivers a real mania — the hard part is not buying. It is selling enough to make gains real. Paper gains are fake money. They become real only when converted into security and optionality.
Small company potential. If the agent system or research workflow becomes useful enough for other people, it may force a product. I do not enjoy managing people. A small team of three to five could work. Anything larger needs a partner who likes operations more than I do.
Returning to a company for a while. PM experience, crypto history, hands-on AI agent work may become more valuable in 2027-2028. Not failure. Just a curve, not an identity crisis.
Chengdu as physical base. Chengdu may offer the mix I want: family stability, lower friction, food, culture, and enough space to keep building online for a global audience.
The black swan. Regulation. Health. Family needs. Crypto failure. Something I cannot model. Liquidity matters. Asset-liability matching matters. Stay in the game first. Then play well.
What I cannot outsource
AI can compress research time. It can remind me of old decisions. It can scan the world while I sleep. It can write, code, summarize, and argue with me.
It cannot choose the life I want. That part is still mine.
The most likely future is a mix: steady path, plus one or two big cycles, plus a possible small product that grows out of my own workflow.
Many people in crypto either go all in until they break, or leave the table after one good round. I want something different.
I want to stay exposed to upside while protecting the base layer of life.
That means systems over moods.
It means selling some dreams when the market is drunk.
It means building before the trigger arrives.
Ten years from now, I do not need the exact scenario to be right.
I need the direction to be right, the family to be safe, and the system to keep improving.
Stay Invested in the Game.