We built an MCP server that did this for Claude about six months ago (Driggsby). We've evolved since then into a standalone financial devtool, but several interesting learnings building something in this space:
- The insights you can get from connecting AI to your finances can actually be pretty cool. I've used it for detecting big charges, optimizing my taxes, investing, cancelling subscriptions, negotiating bills, and building my wife a custom dashboard with all our finances that she looks at daily. If you're someone who has traditionally been "bad at money," in the sense of feeling like you're not managing it closely enough, this kind of product could be good for you.
- The read layer for personal finances is commoditized: all the big players (Muse/Meta, ChatGPT, Grok, Claude, Perplexity) now let you connect your finances through Plaid. They all understandably want this data from you, which can help them serve better ads and predict your behavior. The frontier labs will likely own the readonly portion but leave the riskier parts (like money movement) to third parties. Anthropic interestingly mentioned in their press release they want to hear from "licensed professionals and regulated firms" who want to work with them on this.
- Actually building a great financial AI product that works at scale is harder than you'd expect, because of how diverse and messy financial data is, plus the risk of getting something wrong. This takes a ton of iteration to get right, and often simpler AI features with a clear "this is not perfect" give the biggest bang for the buck. Our transaction classification pipeline that corrects Plaid's broken categorizations using a small LLM has been surprisingly useful.
- If you're like me and are primarily a Claude Code user rather than Claude.ai, this still won't work for you. At Driggsby we optimized more for performance in environments like Claude Code and Codex.
- Financial data is incredibly sensitive, so do consider before you link whether you're comfortable with your AI company having access to all your finances, plus your chat history. I'm hearing from friends in the litigation space how valuable chat histories have become for lawsuits. Incentives will guide action.
Yeah, investing over time in such a way to beat the S&P is actually incredibly challenging (having tried it myself). I’m very skeptical an LLM can do better than that unless it has a very large, expensive firehose of data.
It may be that more mundane analysis ends up being the most useful. For example — for years, we had some money in a money market fund earning basically no interest. It just wasn’t on the radar.
Had even a not very smart LLM nudged us to put the money into a HYSA earlier, we would have made thousands per year in interest.
My wife also recently used our tool to do some simple investing optimization that’s going to save us a few thousand per year on taxes.
Thing like this aren’t sexy but do have value, and don’t take GPT Pro.
This was B2B SaaS for large networks of ambulatory centers to manage/pay their vendors. The banks that were in scope were only the ones they used -- each one of the big names (and around half of them Bank of America).
Can see it being more of an issue if it were a B2C finance app.
Local AI models are getting a lot better. If you have the capability to run them, you could automate this yourself using your own browser automation, actually. It is rather fiddly, as mentioned in the post, but is absolutely doable, and probably the only option, at least for now, where you wouldn't need to provide your credentials to a third party.
Plaid does do screen scraping for smaller banks, but they have agreements for OAuth-based access with most of the largest institutions.
The main reason I do it, at least, is because the insights it provides (as someone who's interested in spending patterns/investing/etc.) are actually quite useful. With even a very basic prompt, I've found things that were otherwise completely missed.
Making it secure is actually quite hard, but that's why we're spending an enormous amount of time thinking about that.
- The insights you can get from connecting AI to your finances can actually be pretty cool. I've used it for detecting big charges, optimizing my taxes, investing, cancelling subscriptions, negotiating bills, and building my wife a custom dashboard with all our finances that she looks at daily. If you're someone who has traditionally been "bad at money," in the sense of feeling like you're not managing it closely enough, this kind of product could be good for you.
- The read layer for personal finances is commoditized: all the big players (Muse/Meta, ChatGPT, Grok, Claude, Perplexity) now let you connect your finances through Plaid. They all understandably want this data from you, which can help them serve better ads and predict your behavior. The frontier labs will likely own the readonly portion but leave the riskier parts (like money movement) to third parties. Anthropic interestingly mentioned in their press release they want to hear from "licensed professionals and regulated firms" who want to work with them on this.
- Actually building a great financial AI product that works at scale is harder than you'd expect, because of how diverse and messy financial data is, plus the risk of getting something wrong. This takes a ton of iteration to get right, and often simpler AI features with a clear "this is not perfect" give the biggest bang for the buck. Our transaction classification pipeline that corrects Plaid's broken categorizations using a small LLM has been surprisingly useful.
- If you're like me and are primarily a Claude Code user rather than Claude.ai, this still won't work for you. At Driggsby we optimized more for performance in environments like Claude Code and Codex.
- Financial data is incredibly sensitive, so do consider before you link whether you're comfortable with your AI company having access to all your finances, plus your chat history. I'm hearing from friends in the litigation space how valuable chat histories have become for lawsuits. Incentives will guide action.
reply