Opey – Your Intelligent and Slightly Geeky API Assistant

This year at TESOBE, we began exploring Large Language Models (LLMs) and especially Chat GPT (4o) in relation to our product. What emerged was a happy collision of APIs and AIs named “Opey”.

The promise of Opey (and similar intelligent agents) is a “non-user interface” that presents its answers dynamically as a result of conversations with bank staff members or customers. Before we dive deeper, let’s take a step back.

Remember two weeks ago – when developers had to scroll through pages of documentation to find the right endpoints and figure out how to use them? Well, Opey, your intelligent and slightly quirky OBP API assistant, can speed up the API discovery process for both newcomers and seasoned OBP API developers alike. 

Modelled on GPT-4o, Opey is a software agent informed by the OBP endpoint documentation and glossary, enabling it to understand natural language questions and respond with well-structured answers. Thus, Opey can suggest combinations of API calls that will satisfy the data and service requirements for various business or technical use cases.

AI Chatbots in Financial Services

Though adoption in financial services is currently limited by challenges and risks, financial institutions are already using Generative AI to automate and simplify processes while modernising customer-facing interfaces. 

Some examples include JPMorgan Chase’s internal LLM acting as a research analyst for internal teams and Wells Fargo’s virtual assistant, which leverages Google Cloud’s AI, helps customers access services via speech or text.

Simulation of Fargo screens

For TESOBE, Opey emerged as a product of our API-driven platform and an interest in current LLM developments. You could say that Opey is an emergent property of a formally documented and tagged API system, pre-processing software and LLMs which can “understand” and synthesise the abilities of those APIs. 

How Opey Helps

The best way to discover Opey is to try it out – just ask it a question related to tasks related to the Open Bank Project APIs and see its response. Opey is currently available as a plugin to OBP API Explorer II, and we expect various flavours of it will be available on the OBP API Portal and the OBP API Manager. While we’ve crafted Opey’s personality for this specific context, its tone and style can be adapted to fit different environments. 

But first, let me share a couple of pleasantly surprising interactions I had when I was testing it out.

One of the first questions I asked, apart from “Hello, are you there?” was “What endpoints can I use to get data about ATMs?” Opey promptly responded with a nicely formatted table of endpoints, but one thing stood out – there was a HEAD endpoint on the list, which surprised me because we generally don’t have HEAD endpoints to our catalogue. 

But then, I remembered that we had added HEAD ATMs for a project with the Mexican regulator, CNBV. In other words, Opey presented the information in a way that even an “expert” in the subject domain was pleasantly surprised.

The second question I asked was “How would I get the phone number of a Customer named Simon Redfern?” Again, I was a little surprised by the answer it gave: It correctly showed how to get all Customers and further explained that we’d have to filter the Customers locally to focus on the record we wanted. I initially wondered, “Why not search by name?” before realising that while we have search-by-attribute endpoints, we don’t have one for Customer Name specifically.

In both cases, Opey was spot on and presented the information well. 

Beyond these surprising results, Opey provides many practical, more obvious benefits that make working with the OBP API smoother. For instance:

 

  • Tailored tutorials. We are all used to providing a getting started guide or generic tutorials, but not all developers will continue down the same path. With Opey, developers can request tailored step by step tutorials for their specific use cases and needs.

  • Endpoint Guidance. Opey helps developers understand different API endpoints, providing descriptions, parameters, and usage examples to ensure smooth integration.
  • Authentication Help. Guides developers through various authentication methods, recommending the easiest and most secure options based on their needs.
  • Glossary Assistance. Decodes complex terms used by your API, providing examples and making it easier for developers to understand and use the API effectively.
  • In depth knowledge. Opey has the ability to reveal details that are not obvious, even for the core API team.
  • Free up support: Not all banks can have a dedicated team to provide 24/7 support and guidance to developers. An intelligent, always-available assistant can reduce the burden on technical support teams.

As front end developers we are pretty used to providing one type of Graphical User Interface for everyone and being obsessed with the details down to a few pixels.

The promise of Opey and similar agents is a new type of user interface that is much more fluid than the one we’re used to. It won’t be a fixed thing. It could morph into something completely different depending on the type of data you’re displaying.

Agents for the Future

At the time of writing, Opey understands OBP APIs and can reason about how to use them, how to combine them and how to authenticate. In the near future, Opey will become an AI agent that not only understands APIs but also acts on behalf of the user with some of their permissions. 

Instead of asking a general question: “How would I get a specific customer?”, imagine asking, “Please tell me how our customer Simon Redfern differs from our average customer” and getting an instant, detailed response after calling multiple endpoints and combining the results.  

To do this, Opey will need permissions to access the relevant APIs, which it will manage using OBP’s fine-grained and comprehensive consent mechanisms—broader than those typically seen in UK or Berlin Group PSD2 standards. This positions Opey as a multi stakeholder tool – usable by retail customers, API managers, and bank teams – each leveraging a subset of their API permissions, which can be controlled and revoked at any time.

Opey might be considered a framework that can serve the disparate user groups of a bank running OBP. It’s an app that reinvents itself based on who’s using it and the questions being asked.

Beyond internal data, Opey could also extract relevant information from the web, opening new possibilities. For instance, Opey could gather market trends or conduct competitive analyses, providing a richer context for decision-making. The AI’s ability to synthesise data almost instantly is a game-changer.

Test it out!

Without further ado, it’s time for you to test Opey.

https://apiexplorer-ii-sandbox.openbankproject.com/

We invite developers to provide feedback and suggestions as we shape Opey’s future. Your input is invaluable in guiding its evolution to meet your needs.

Conclusion

Opey is not just an intelligent assistant but a geeky sidekick with a fun personality that makes technical support feel like a friendly conversation. With a blend of humour and expertise, Opey transforms complex tasks into manageable ones, revealing often-overlooked details. It’s going to become an indispensable tool for developers navigating the OBP API.

 

Cheers!

Simon