The real return on AI in finance is not just faster work but novel insights through deeper understanding, broader coverage and smarter decisions.

In financial services, the promise of AI has long been measured in time saved. But shaving 20 minutes off a task is not the full story. The real opportunity lies in how that time gets reinvested and what impact it drives downstream. 

I discussed this in detail with James King on the IA Talks AI podcast, along with Rufus Saunders, who is responsible for Business Development and Commercial Strategy at Finster. What follows is a look at what productivity actually means when AI meets capital markets.

You can listen to the full episode here.

Time saved is not productivity

Yes, AI can help analysts complete a draft in 20 minutes instead of two hours. But that alone is not the win. The question is: what happens to the rest of the hour and forty? If it’s spent doing the same task faster, you’ve only shifted pace. If it’s used to cover more names, go deeper into the supply chain, or surface new red flags, then the ROI starts to compound.

In this sense, “productivity” is not about fewer hours. It is about better decisions. More ideas tested. Bad ideas killed quickly. Less time wasted going down the wrong path.

Second-order impacts are where AI pays off

This is what we call deep productivity. It is when saved time leads to:

• Smarter decisions, made sooner

• Wider coverage without more analysts

• Faster iteration of ideas

• More differentiated work 

These outcomes matter more than measuring how long something took. They show up in metrics like AUM per analyst or coverage per desk.

Financial services professionals already know where they add value

Many already use ChatGPT on the side. They understand what’s repetitive and what’s high impact. Finster turns that instinct into a safe, auditable system. It automates the 70 to 80 percent of work that machines can do and gets out of the way for the rest.

That means they get to their “magic” faster. They spend more time applying judgment, sector knowledge, and intuition, rather than reformatting decks or searching PDFs.

The right answer might be ‘I don’t know’

In finance, being wrong is worse than being slow. That’s why Finster is designed to say “I don’t know” when the data doesn’t support a conclusion. General-purpose models try to answer everything. We don’t. If a source can’t be found, Finster says so.

This matters because hallucinations aren’t just an inconvenience here. They are a liability. A wrong revenue figure in a comp set can cause real harm. Saying “I don’t know” is sometimes the smartest thing an AI can do.

ROI must include the people using the AI

Another challenge: many senior leaders don’t always know what their analysts do hour to hour. That makes it hard to quantify what’s gained when work gets easier.

Finster sells into workflows, not features. We show how an initiation note gets done today and how 70 percent of it can be automated, with full traceability. That lets the PM or analyst focus on the 30 percent that actually generates returns.

Adoption requires change management, not hype

Firms do not move because of tech hype. They move when AI helps analysts do more with the same headcount and when compliance is on board. Finster was built for both.

We don’t assume every organisation is ready. We run live trials, measure output, and help teams prove the case internally. That means being realistic about the complexity of the work and showing real gains without disruption.

AI won’t replace analysts. It will amplify them.

Firms that get this right will be able to save time and reallocate it to higher-impact work. They’ll grow without growing headcount. And they’ll bring better research to market, faster.

That’s what deep productivity looks like, and that is the real ROI of AI in finance.