Finastra Mashup Challenge
The Fintech Lego
At Finastra Labs we are experimenting with 0 day technologies every day. Naturally we come up with plenty of ideas during this journey, but we don’t have the luxury to build all of them. We are not limited to soon to be productisable concepts, but those without a high potential to our future markets get easily under prioritised and sunk finally in the Backlog Ocean. This led us to organize an internal event so we can satisfy our desire to play and learn even from our wildest ideas.
At first a catalogue of publicly available services, APIs, SDKs has been collected and published one week before the event, so everybody could become familiar with the technologies. Then we split into 3 groups, each group with product manager, designer and developer members to build balanced teams. We had one rule, and one rule only that the teams were to come up with a project idea that included at least 2 technologies from the catalog. The teams’ creativity was the limit for every other aspects.
Here are the list of services that the teams could apply to their projects:
- Natural language understanding (Wit.ai, Api.ai, Recast.ai, Luis.ai)
- Infrastructure (BeepBoop, Firebase, Microsoft Bot Framework)
- Data/account integration (Plaid)
- Image recognition (CloudSight, Google Vision)
- User identity validation (EyeVerify, Jumio)
- Machine learning (Google Prediction API, Microsoft Azure ML Studio)
- IoT/Physical Web (Estimote, LittleBits, Web Bluetooth API)
- All-in-One Suites (Algorithmia, IBM Watson, Microsoft Cognitive Services)
- Service Automation (IFTTT, Zapier)
Most of them are pretty popular, but should you have deeper interested on any of those you can find our catalogue here with brief descriptions and links to details. All of them have free tier or trial period, and easy to get started with, but feel free to reach out to us if we can help.
We gave the Mashup Challenge name for our event, as mashup means to use 2 or more services and combine them to create a new service. The real challenge was that the teams had 1 day only to build a working and meaningful prototype from their ideas.
Our #1 goal was to validate a Fintech Lego model, and check how rapidly can we prototype a new idea leveraging publicly available APIs. It can demonstrate that on top of our solid platforms and products we can build an ecosystem for our clients, bring in innovative features and showcase them in days or even hours. In this challenge there were 6 hours for the participants to design and implement their prototype once the teams have been formed and agreed on the idea. Let’s see what did we manage in this timeframe!
Team#1 created a funny language learning app. The user can learn a new word each day. The app asks for the word and extract the user’s answer via speech-to-text API then understand it with the wit.ai NLU API. When it’s done the app triggers a funny LittleBits creature. The 21st century edition of the famous toilet paper roll sudoku concept and you don’t need diarrhoea to master a new language.
The second team created a chatbot for Slack that is capable to recognize bug reports and feature requests in the conversations. It pushes through every message on API.ai’s NLU engine that is able to detect the intention of the sentences. Only a few example message was enough to train the model and get remarkable efficiency. It’s been integrated to our project backlogs boards in Trello with an IFTTT (“if this than that”) applet. It’s still live and some team members are using it as a work efficiency booster.
Last but not least, the third team created a Facebook Messenger chatbot that is able to control your smart home. You can make a natural conversation with the bot and get information whether you locked the front door or what value the thermostat has been set to. Also you can switch on/off the lights on each room, lock/unlock the door, open the garage door, change the thermostat, etc. Only those would call it silly who never returned with doubt whether the door was locked!
During these hours we had a lot of fun, freed up some creative energy and learned a lot about the technologies we used. On one hand it was a good practice for prototyping in extremely short amount of time, but on other hand we managed to validate that we can very easily lego any new ideas on top of our platforms to come up with the bests of the bespoken and the out of the box worlds.
Interested in details? Reach out to us