Key insights
- The author suggests AI adoption in finance will be slower than expected due to 'adoption inertia' (sales cycles, compliance, risk). This creates a short-term opportunity for small investment teams and builders to leverage AI for outsized returns before best practices are established and larger players catch up. This could lead to short-term productivity gains but no significant market impact.
I spent the last few months testing ChatGPT and Anthropic’s Excel add-ons for investment workflows. And my main takeaway is that while these tools are still rough around the edges, I'm starting to see their potential. The feeling I get is that AI for finance is starting to look a bit like coding did a few years ago.
This got me thinking. Adoption for coding took off because LLMs like Sonnet 3.5 reached an inflection point and only got better from there. Right now, model capability is no longer the bottleneck.
Instead, I believe the bottlenecks is what I all "adoption inertia", which are things like:
- Longer sales cycles for enterprise deployment (especially for legacy industries) * Compliance / security / privacy * Risk management: companies may not want non-deterministic AI into customer-facing / mission critical workflows * Internal pushback from parts of leadership or individual contributors
Therefore, not every industry will adopt AI at the same speed. There are a couple downstream implications for this.
For investors: because the best practices for using AI haven't quite yet been established, so small investment teams can use AI in novel ways that can give them the capabilities of much larger platforms. We already see some of the folks on this subreddit post interesting tools
For builders: small teams can use AI in interesting ways to generate revenue extremely quickly. These don't even have to be software companies (one example here is two brothers who used AI to help them sell GLP-1s at scale).
The edge probably won’t last forever. So it really feels to me that there's a small window of opportunity for small, scrappy teams (that don't have significant capital or proprietary tech / distribution) to generate outsized returns.
I wrote a full post about it here.