AI is Changing Customer Interactions. Will Your Bank Join the Conversation?
By Chris Griffin, Co-Founder at Narmi
Simple Subscribe
Subscribe Now!
Who has the answers to AI in this industry?
As we’ve said before, an incremental approach is the only one that banking – with its web of stakeholders, complex data, and the strength of its regulatory apparatus – is likely to pursue when it comes to artificial intelligence. The technologies that succeed will be developed with intention, taking into account that the financial backdrop of our lives needs to function reliably.
Even before AI, the lack of fulsome and reliable integrations plagued the banking world – both for financial institutions themselves and the consumers and businesses that rely on them. Banks and credit unions that have invested in building and consuming reliable APIs have the appropriate foundation to invest in generative AI integrations. So what AI investments should that financial institution make today?
Enter the Model Context Protocol (MCP): a new way for financial institutions to securely connect their users’ banking data to LLMs like Claude and ChatGPT. While MCP is not the only AI investment a financial institution should make, it is perhaps the most foundational. Acting essentially as a standardized API purpose built for AI, think of MCP as a baselayer for all of the permissioned user data connections your financial institution wishes to empower.
Want to read more like this? Check out Narmi’s content portal on The Financial Brand: Be Where Banking is Going
MCP in Action
What that looks like: imagine allowing an end user to log on to their personal instance of an LLM like Claude and ask it something like,
“What are my two highest categories of discretionary spending this month?”
And for the LLM to return an answer like:
Restaurants -$140.24 (14.2%)
- 7 transactions at Chipotle, KFC, Chipotle
- Average $20.03 per meal
Entertainment -$98.17 (9.8%)
- 9 transactions at Apple Music, Netflix, HBO Max, Criterion Collection, Disney+, Paramount+, YouTube Premium, Spotify, Hulu
- Average $10.90 per transaction
Total spending in these categories: $238.41
- The majority of your discretionary spending this month so far has gone towards restaurants, with a significant portion of entertainment spend going towards streaming subscriptions. These comprise a 23.4% share of your discretionary spending this month so far.
Noticing a trend, your end user then asks: “That’s a lot of money to spend on subscriptions. Give me a ranking of all of my subscription purchases this month, ordered from most expensive to least,” creating a conversational financial planning session.
And the use cases aren’t just limited to personal finance: imagine a business owner asking, “How much did my business spend on software and marketing last month, and how does that compare to revenue?”
And like most foundational technologies, once the infrastructure is in place, the possibilities extend past just the initial use cases. It’s not a matter of if, but when.
The Future of the Internet
When we think about the average user’s experience, it’s clear that eyeballs are shifting from the search engines and app stores that have historically mediated our relationship with the internet to LLMs. Gone are a list of SEO-optimized links in response to a Google search – a growing cohort is turning to generative AI for their daily needs. With this shift, we’ll see LLM clients like Claude, ChatGPT, and Gemini hold significant power over what users see on their platforms, creating the potential for a future where the internet is sliced up into different marketplaces maintained by LLM providers.
And the technology behind these marketplaces? MCP.
As we see more power being given to LLMs as the funnel to the larger internet, they’ll have the ability to dictate the rules of the market – including which MCP integrations attain priority. Companies as established as Stripe, Shopify, and Figma have already built their own MCP servers, likely in anticipation of having prime position within the new MCP marketplaces. For example, Anthropic has already rolled out a marketplace offering “Connectors” to various services, allowing users to integrate and use its Claude LLM as a platform to:
- Search and update your workspace (Notion)
- Organize your calendar (Google Calendar)
- Manage technical team projects and workflows (Linear)
- Oversee payments activity (PayPal)
Now, imagine your financial institution on this list. We’ve established that an MCP server can deliver a huge amount of value for your financial institution’s customers/members. But an essential driver of the technology’s appeal for – and perhaps, especially for – community banks and credit unions is that it can act as a new distribution channel to reach your customers.
In a future where an LLM could serve as a user’s primary “homepage” for internet interactions, being visible in these emerging marketplaces could be akin to achieving a high-ranking search result. This could bring substantial attention to your institution’s offerings and showcase its forward-thinking functionalities.
While the current market is in flux, it’s crucial not to cede the development of MCPs solely to the major players. By moving quickly to develop your own MCP, your financial institution could position itself to be among the early participants in these marketplaces. This could make your institution competitive against larger banks on a channel that has the potential to be more democratic than the SEO-driven search internet.
Through the MCP market, financial institutions could have an opportunity to make the face of their brand an inherently sticky functionality that meets organic business and consumer demands. It’s an opportunity for local financial institutions to make inroads outside of word-of-mouth; to truly be where banking is going.
Leveling the Playing Field
Community banks and credit unions need to innovate in the direction of consumer and business expectations. The notion of “table stakes” has become a moving target, and it’s more important than ever to dedicate resources to where eyeballs are gravitating.
The Model Context Protocol is poised to be a pivotal development for financial institutions. We’re only just beginning to talk about its potential for ubiquity, but we could be looking at a watershed moment in making the vast majority of financial institutions more competitive, breaking down traditional barriers and allowing them to pursue their mission – connecting with people across the country, and elevating their financial lives.
