GUIDES
Finance AI Week đź“š
Anthropic and Perplexity both dropped major finance updates on back-to-back days. Here’s what each one actually does, what plan you need, and three things to.
What Just Happened
On May 4, 2026, Perplexity launched Computer for Professional Finance — 35 ready-made finance workflows pulling live data from Morningstar, PitchBook, Daloopa, and Carbon Arc.
One day later, on May 5, Anthropic announced 10 finance agents plus full Microsoft 365 integration (Excel, Word, PowerPoint). Dario Amodei and JPMorgan CEO Jamie Dimon shared a stage in NYC — the first time ever — to announce it.
Finance is one of the most cautious industries in the white-collar world. Two of the biggest AI companies just rebuilt it. That tells you everything about where every other industry is going next.
Anthropic vs Perplexity
They overlap, but they’re for different jobs. Here’s the cleanest way to think about it:
Anthropic
10 Finance Agents + Microsoft 365
Best for: doing the work itself — building models, drafting decks, writing memos.
Plan: any paid Claude plan (Pro, Max, Team, Enterprise).
Where it lives: inside Excel, Word, and PowerPoint as add-ins.
Perplexity
35 Finance Workflows
Best for: getting the data — tearsheets, comps, market research.
Plan: Perplexity Max ($200/mo) for individuals.
Where it lives: a workspace that connects to Morningstar and PitchBook.
Anthropic’s 10 agents: Pitch Builder, Meeting Preparer, Earnings Reviewer, Model Builder, Market Researcher, Valuation Reviewer, GL Reconciler, Month-End Closer, Statement Auditor, KYC Screener.
Perplexity’s workflow categories: Real Estate, Private Equity, Public Equities, Hedge Funds, Asset Management, Wealth Management, Investment Banking, Insurance, Credit, Corporate Finance.
Bloomberg vs Perplexity
A Bloomberg Terminal seat costs about $32,000 per year. Perplexity Max is $2,400 per year. Roughly 13Ă— cheaper.
It doesn’t replace every Bloomberg use case. Bloomberg still wins on real-time trading data and deep institutional licensing. But for research, tearsheets, comps, and screener-style work? The math just changed for everyone who isn’t a top-tier hedge fund.
3 Things You Can Actually Do
You don’t need to work in finance to use any of this. Pick the one that maps to your job and try it this week.
Install the Claude Excel add-in (any paid plan). Paste in a public 10-K link. Ask: “Build a 3-statement model with sensitivity tabs from this filing. Show me the working assumptions in a separate sheet I can edit.” Then ask Claude to draft the matching PowerPoint summary that auto-updates if you change an assumption.
In Perplexity Max, run the Company Tearsheet workflow. You get a one-page report with valuation, peer comps, earnings history, and analyst ratings in under 10 minutes. Export as PDF and you have a deliverable. This used to take a junior analyst half a day.
In Claude Cowork on a paid plan, install the Pitch Builder plugin. Give it a public ticker. Ask: “Build a comparable-company analysis and a draft pitch deck for [TICKER]. Tell me which slides you need more data on before I show this to anyone.” The last sentence is the trick — it forces Claude to flag where it’s guessing.
The real unlock of AI agents is not full autonomy, but human-in-the-loop systems design. Structure your agent inputs with explicit constraints, negative prompts, and automated test checkpoints. When building tools, keep token consumption lean by caching system prompts and isolating tasks into specialized sub-agents.