The AI Analyst Is Here: How Claude Is Quietly Entering Global Finance

In July, something remarkable happened on a public stage. Anthropic, the AI research lab behind Claude, demonstrated its first-ever AI financial analyst.

In just two minutes, the AI:

This wasn’t a toy demo. This was the real work of a junior investment analyst — done in real time by an AI agent.And now, this same financial AI is moving from demos into the real world. Anthropic has begun rolling it out to two of the largest sovereign wealth funds on Earth:

At the same time, Claude has also been deployed across Deloitte’s 450,000 global employees. This is not a future promise. It is happening now.

Instead of waiting to build “general” artificial intelligence for every industry at once, Anthropic made a strategic choice: start with one difficult vertical and go deep. They chose finance. Why?

In programming, small mistakes break software. In finance, small mistakes can cost millions or billions. The skills required — logic, data structure, mathematical reasoning, tracing assumptions — are exactly what AI systems trained on coding tasks already do well. Accounting, financial modeling, and risk assessment are really just “domain-specific languages” — similar to coding, but expressed in numbers instead of code. So if an AI can write, debug, and reason over software, it turns out it can also reason over balance sheets and valuation models. That is why finance became Anthropic’s first true vertical AI initiative.

Planning your estate might not sound like the most exciting thing on your to-do list, but trust us, it’s worth it. And with The Investor’s Guide to Estate Planning, preparing isn’t as daunting as it may seem.

Inside, you’ll find {straightforward advice} on tackling key documents to clearly spell out your wishes.

Plus, there’s help for having those all-important family conversations about your financial legacy to make sure everyone’s on the same page (and avoid negative future surprises).

Why leave things to chance when you can take control? Explore ways to start, review or refine your estate plan today with The Investor’s Guide to Estate Planning.

Download your free guide now!

Anthropic describes their agent workflow using three simple verbs:

Gather information from filings, databases, reports, and public records.

Run calculations, build models, perform sensitivity testing, spot trends and anomalies.

Produce usable work outputs:

This mirrors exactly how real analysts work.

The industry benchmark for finance retrieval tasks is called “Finest Agent.”
It tests whether models can answer questions like:

This is mostly entry-level research work — not deep modeling yet.

On this benchmark:

That might not sound impressive — but here’s the important lens: These are the worst these tools will ever be. AI has only been focused on finance as a vertical for three months — and in that short time alone, results jumped by five full percentage points. This shows huge room for rapid improvement.

One of the clearest signs of Claude’s progress comes from modeling tasks in Excel.

Example:

Claude:

This used to require:

What took analysts minutes or hours now happens in seconds.

$200B under management.
~200 employees total.

BCI faces a classic problem:

Claude is helping them:

Instead of static spreadsheets:

Senior managing directors now explore data themselves, rather than waiting for reports to be rebuilt.

$2 trillion assets under management. Thousands of staff. Heavy internal engineering teams. Norges built:

Portfolio managers now:

Claude becomes a search-and-analysis copilot embedded directly into their internal tools.

Deloitte rolled Claude out company-wide across:

Consultants now use AI across all three verbs:

The scale here matters: Nearly half a million professionals now have daily access to financial-grade AI.

Good news:
You don’t need a perfect data lake to start. Claude can already:

However, for advanced workflows:

Even traditionally slow data vendors like:

have now launched fully functional MCP integrations — something that used to take years. AI adoption is forcing data infrastructure modernization across finance.

There is no heavy corporate “design partner program.” Instead:

Customers often discover new uses that the AI teams never predicted. Because AI remains:

The best use cases emerge directly from real-world experimentation.

The industry today is strongest at: Research & Retrieval

The next major waves:

The ultimate goal: AI co-workers that take on full projects end-to-end — not just isolated tasks.

Anthropic’s view is simple:

Their ambition: Make Claude the backbone intelligence layer for financial services — regardless of what software interfaces companies prefer.

Some organizations will use Claude directly. Others will:

As long as Claude is powering the intelligence underneath — Anthropic is winning.

The biggest problem in enterprise AI adoption isn’t capability — it’s behavior change.

People don’t want to learn new apps. So Anthropic is placing Claude into existing tools:

Claude appears as a co-worker where people already work — eliminating friction. You don’t “learn AI.” AI becomes part of your daily workflow without disruption.

Shocking statistic:

Adoption is accelerating, but security approval cycles and compliance still take time. Industry expectation:

Enterprise software adoption has never moved this fast in history.

July’s two-minute demo marked more than a cool tech reveal. It marked the beginning of a real shift:

AI isn’t just answering questions anymore — it’s doing financial work.

Today, the AI analyst is an assistant. Tomorrow, it becomes a teammate. We are witnessing the first real verticalization of AI. And finance — the world’s economic engine — is leading the way.