Looking at the Bankbot Landscape

Will Thomsen
Chatbots Magazine
Published in
5 min readNov 2, 2016

--

A few days ago, Keith Armstrong did a nice writeup about the proliferation of chatbots in money and banking. In his post, Keith lists a few personal finance chatbots with funding.

In developing Ledgee, my personal finance chatbot, I‘m keeping a close eye on these names. Here’s a more complete list of players both big and small that are on my radar:

  • Ledgee — my project. Multi-platform personal finance assistant. Available on SMS/Messenger/Alexa. Focused on evolving a personalized experience that becomes more useful with each conversation
  • Abe — originally a slack bot for teams, but now seemingly more focused on personal finance
  • CapitalOne — Alexa skill offering basic account management
  • Cleo — UK-based assistant. Started on SMS but now focused on Messenger
  • Digit — automated savings app using a conversational interface
  • Dyme — white label Messenger/SMS assistant
  • Erica (Bank of America) — voice & text. Making nice attempts at using predictive analytics to drive actionable advice
  • Finie (Clinc) — University of Michigan upstart. Showcases the lab’s Lucida project, focused on superior NLP. Aiming for white label integrations
  • Finn — white label multi-platform assistant
  • FIVE (Best Innovation Group) — white label Alexa skill for credit unions
  • Kasisto — Messenger assistant. Originally launched standalone, but recently announced partnership with Mastercard
  • Trim — subscription management bot. Has support for some commands but no NLP
  • Penny — conversational personal finance app. Uses a button/command tree in lieu of NLP. Actually has transaction categorization (which, unfortunately, is still a must)
  • Xobi — app with natural language search. Good visuals

I’ve had a chance to try a few. Here are some takeaways that are driving Ledgee’s development.

Digit is focused on automated savings, but is evolving its intelligent conversation capabilities. It uses a chat UI to handle transactions, but also supports other basic commands.

When it comes to personal finance apps, action is paramount. The days of read-only apps are fading and chatbots won’t be an exception. I call this the “don’t just stand there, do something!” requirement. Digit gets this right, and goes a step further by automating the action. Todays personal finance apps need to offer smart and seamlessly actionable advice.

Digit is a good place for someone to start a savings habit, but it isn’t fundamental. I’m looking forward to the update that transforms Digit from the fun app that sacks away a little money to the foundational app that implements a more robust savings plan.

Abe originally positioned itself for businesses/groups. It was a Slack bot meant to help teams handle finances. Recently Abe expanded to SMS and is now taking aim at personal finance.

In trying Abe, one area that impressed me was its feedback mechanism. In the conversation, Abe prompts the user to use a hashtag along with the feedback. There’s no extra form or pop up. It’s done right in the conversation. I don’t know if this is something Abe imagined, but it’s very well done.

This has me thinking about other ways hashtags could be used within chatbots (#human comes to mind). Would everyone would benefit from a universal list of hashtags implemented in every bot? Or are pre-canned commands exactly the sort of thing we’re trying to avoid?

I still think there’s value to the business/team focused banking chatbot and I hope Abe continues to explore it. While I’m less convinced at the corporate level, it makes a lot of sense for smaller organizations. Nonprofits, volunteer groups, extracurriculars, meetups and open source projects would certainly benefit from chatbots assisting in areas such as:

  • donations
  • expense reports
  • account/budget management
  • financial transparency

Trim is a personal finance chatbot operating in the newer subscription management space. This places it in direct competition with a few other young startups (most notably, TrueBill). Trim also has commands to get quick information about your accounts.

Like Digit, Trim fulfills my “don’t just stand there, do something!” requirement. It offers to cancel unwanted subscriptions on behalf of the user. I’m on the fence regarding subscription management as a standalone business, but it’s certainly a strong launchpad.

When I used Trim, it lacked NLP, opting instead for explicit commands. This works better in Messenger’s rich conversation platform, but it still feels like a miss not to have both. Trim also believes I have a subscription to a nearby sandwich shop, but I suppose that’s on me.

Penny is a personal finance chatbot living inside an app. Penny seems most focused on budgeting.

Inside the conversation, users interact with Penny by tapping through a tree of buttons. This is a double-edged sword. Discoverability is a breeze, but it loses much of the “infinitely wide top level” benefits of a chat interface. I think the best chatbots will blend both methods.

Penny is well done. Transaction categorization is easy and the fact that it even exists places Penny ahead of many competitors. It also shines in the details, for example by mimicking the “user is typing something” indicator bubbles while fetching the answer to your inquiry.

After launching a standalone bot this summer, Kasisto is now partnering with Mastercard to offer Mastercard KAI.

One interesting area for Kai (at least in the standalone app) is its conversational budgeting. During a conversation, it asks users if they want to set a monthly budget for a certain spending category. If so, it will then send notifications when you overspend in that category.

To succeed, I believe chatbots must build a personalized experience. Notifications aren’t groundbreaking, but they’re more compelling when suggested after a user indicates interest. This is headway towards personalization and I’m excited to see how they build on it.

That’s it. It’s a fast-growing space, and I’m excited to see how each of these companies evolves the conversational interface. If you know of other names that belong on this list, please let me know!

--

--