On AI Copilots for Finance

For a while, I have been interested about how AI is reorganizing knowledge work in finance. It all started in my sophomore year when I got a tour of the Farsight office. At this point, the market feels like it is settling into three layers: research, diligence, and execution. Research includes companies like Rogo, AlphaSense, and Bloomberg AI. Diligence includes companies like Hebbia and Latch. Execution, where analysts actually turn work into deliverables, is where I believe Farsight stands out.

Farsight is the clearest example of execution automation I have seen. The product sits inside Excel and PowerPoint and focuses on the exact outputs analysts produce every day. It learns a firm’s templates and formatting, reduces hours of updates and revisions to minutes, and integrates into workflows without asking anyone to change their habits. Since analysts already spend their lives in Excel and PowerPoint, this approach fits how they work especially vs. other tools I've seen that don't sit within (e.g., Nummo).

The execution layer feels especially compelling because it sits closest to the final deliverable, which is more the moment of value inside a bank or an investment firm. It creates switching cost and it is difficult to copy because structured deliverables and deep Office automation require both technical depth and workflow understanding.

Farsight’s performance supports this view. Revenue grew 10x in 2024 and customer count grew 5x. The recent $16m Series A led by SignalFire also adds some great validation.Varun, part of the SignalFire team that led this investment, is someone I've previously tried to talk to. I am also hearing mixed feedback on Rogo (a click more traction than Farsight) from friends who expected stronger quality. If research layer tools do not meet expectations or do not demonstrate differentiated value, I can see budgets shifting toward the execution layer. I think it is easier to defend this spend because it ties directly to output quality and speed.

Looking forward, I think Farsight will capture even more budget allocation as it continues to outcompete on execution. Even if new entrants come into this layer, deep integration, output quality, and innovation speed matter more than anything else. If Farsight keeps shipping at its current pace, it strengthens its position as the default tool analysts rely on.

It will be useful to track proof of concept conversions and the path from a small pilot to broader rollout. The product delivers its value immediately, so conversion rates should benefit from how visible the impact is. Once a few people on a desk use it, it becomes clear how it saves time and cleans up output quality. Then, the GTM motion is natural expansion to other teams within a firm, as well as FOMO-driven intake from banks/firms' competitors.

During my conversation with an investor at NEA, he pointed out that seat expansion within existing customers is a metric worth watching. I think that's an important reminder that horizontal adoption across desks or groups can be just as meaningful as landing new accounts. Actually, it probably is more useful ('FOMO-adjusted').

That NEA investor did bring up a few more areas to consider (i.e., how difficult is replicating search vs execution?); though, I keep coming back to the strength of the execution layer. It sits closest to value creation, aligns cleanly with budgets, and benefits from visible wins during POCs and natural seat expansion. If Farsight continues to innovate quickly, it has a path to grow from a focused tool into something that feels much closer to a platform inside the modern finance workflow.