If you know me, you know I’d rather be writing about a $38.5 million Ferrari or the latest Concours d’Elegance. But here we are. My most viral tweet ever has nothing to do with cars. It’s about AI and wealth management. Of course it is.
A couple of weeks ago, Anthropic released their Wealth Management plugin for Claude Cowork. As someone who has spent 17 years building MProfit, a wealth tracking and analytics platform for family offices and advisors, I obviously had to dig in. I put together a 12-part thread walking through the plugin, what it does, how to use it, what the inputs and outputs look like, and I shared all the PDFs involved. I expected a few hundred likes from the usual fintech crowd. Instead, it blew up. 2.2 million views. 11,000 reposts. 23,000 likes.
My phone didn’t stop buzzing for two days.
I think the reason it resonated is because the thread wasn’t hype. It was a practitioner’s take. I’ve been in this industry long enough to know what a portfolio review actually involves. The time it takes. The nuance. The back and forth between an analyst and an advisor before anything gets presented to a client. So when I showed what Claude’s plugin could do with a sample MProfit report, people paid attention. Not because AI is new or shiny, but because the output was genuinely useful. And fast. Uncomfortably fast.
This wasn’t the first time I tested LLMs on wealth management workflows. Back in February, I had already run a head-to-head comparison between ChatGPT, Gemini, and Claude using MProfit’s Advanced Performance Review report. I uploaded the same PDF to all three and asked each one to review it from the lens of a wealth manager preparing a report for a family office principal. ChatGPT was fast but bare bones. Gemini looked polished. Claude made me sit up from my chair. The PowerPoint it generated through the Cowork plugin would have taken an analyst days to create. I wrote about all of this on the blog.
But the Wealth Management plugin thread took it further. It wasn’t just about what an LLM can do with a single file anymore. It was about what happens when you give it specialized tooling, connectors to real data sources, and workflows designed for how wealth managers actually work. That’s what the plugin does. Portfolio analysis, drift detection, tax exposure, rebalancing recommendations at scale. Anthropic even partnered with firms like LPL Financial to develop this stuff.
Now, before anyone thinks I’m saying go upload your client data to Claude tomorrow, let me be clear: I’m not. Most of the family offices I work with are very cautious about putting anything into an LLM. And they should be. Data privacy is a real concern, and the question of whether these models use your data for training is one every firm needs to answer for themselves before jumping in.
What I am saying is that the direction is clear. What used to take an analyst one to ten days can now happen in minutes. The quality isn’t perfect, but it’s getting better at a pace that should make anyone in financial services sit up and think about what their job looks like in two years. The thread went viral because people felt that in their gut.
The funny thing is, I’ve been tweeting about cars for years. Ferraris in Mumbai, Bugattis at auctions, barn finds in Rajasthan. Hundreds of car posts. And the one that takes off is about a wealth management AI plugin. I can’t even be mad about it. It’s the intersection of everything I’ve been doing for the last two decades: finance, technology, and trying to figure out where this industry is headed next.
If you haven’t seen the thread, go check it out. And if you’re in the wealth management space and still on the sidelines when it comes to AI, I’d say the sidelines are getting smaller by the day.
You can review the output from the Claude plugin below:
The Wealth Management Client Review doc
The Rs. 20 Cr. investment proposal
The PowerPoint preso for the principles of the family
celestri - virtual home of manish jain 