A lot of people still think AI's impact on productivity is overstated. I disagree. For knowledge workers, the impact is arriving fast.
Coding agent is solved
The technical problem is done. What remains is execution — how fast companies adopt it, how serious they are about rolling it out, and how quickly the rough edges get smoothed. These are solvable problems, not research problems.
This means coding agents will be fully mainstream this year. The technology is through. Everything else is adoption.
What comes next
On top of coding agents, every knowledge industry — legal, finance, medical — has massive room to expand. The technical barriers are not that high anymore. The real work is mapping domain knowledge onto the technology and building customized adoption paths.
Using coding agents as the foundational tool is a powerful idea. Build the tool layer first, then spread across finance and every other industry. The speed will be fast. Within a quarter, we could see technically mature solutions across multiple domains, followed by large-scale rollout.
The knowledge industry endgame
Replacement — or at least deep restructuring — of knowledge work in the highest-value industries could settle this year. That's why I'm spending more time at the intersection of finance and agents. The timing is right.
And this isn't about back-office operational workflows — accounting automation, trade reconciliation. Those are small bets. The moment is right to look at where the real value sits in capital markets and figure out what can be built there.
This connects to projects I've been working on — AI-powered financial analysis, derivatives structuring, valuation agents. The timing has arrived.