FinBot

Project Owner

  • Wayne Andrews (Chief Financial Officer)
  • Stephen Blunt (Financial Controller)
  • Christine Tsintarakis (Associate Director, Engagement & Business Operations)

Project Credits

  • Jim Cook (ICT, TechLab)
  • Stephen Blunt (Financial Controller)
  • Christine Tsintarakis (Associate Director, Engagement & Business Operations)
  • Christopher Milevskiy (Senior Support Analyst, ICT)
  • Dawson Britto
  • Lee Borlace
  • Reto Freihofer
  • Mathew Menzies
  • Paul Hutchings
  • Julia Hamilton (Finance Graduate)
  • Varun Kalia (FSC Operations Manager)

Start Date

18th August 2017

Status

Pilot Delivered February, 2018 Expected Production Environment delivery is early April 2018 (pending ASRB)

Problem Statement

Cognitive services are a viable tool set to replace low value tasks with automation. Many Finance 1st level support tasks fall into this category.

Final Brief

A proof of concept was developed by Finance, ICT Tech Lab, and 3rd party vendors (Ignia & Microsoft) to use Oracle PeopleSoft Accounts Payable data to explore the detail available through a bot. Customer facing use cases were defined through a series of stakeholder engagement workshops, led by Microsoft and Ignia, with participation from Finance, TechLab and Faculty users. The primary use cases identified during these workshops extended to the following functions exclusively:

  • Purchase order status
  • Finance FAQ
  • Supplier identification
  • Invoice status

The bot was written by Ignia and the platform was deployed under the TechLab’s testing environment. The bot was deployed via the Microsoft Teams channel

To productionize the proof of concept and provide an interface within MS Teams and the Intranet, additional code for a web front end was developed.

Note: this is the first of many planned chat-bot solutions with similar functionality currently planned for the Library, Sydney Student, ICT, HR and other parts of Operations.

Challenges & Learnings

  • C1. While the capabilities exist to deliver exploratory agile projects within ICT, the impetus to support them does not.
  • L1. If the project is managed externally with ICT engagement, a favourable outcome can still be achieved.
  • L2. Sometimes you have to build technology before you can explain it to people. Cognitive fits into the category very well, and keeping a library of easily deployable and demonstrable cognitive Service based tools will help with future projects of this nature.

Languages / Framework

C#, Azure, Peoplesoft, Luis.ai, QnAmaker.ai

Links to Resources

The bot is currently deployed in teams. If you have a University UniKey you can search for the following token key to talk to the Dev bot: 65dd2fd6-7457-42bf-8edc-d55d6c977401



Tags: