We’ve just launched an innovative new chatbot that is designed to help staff in an IT consultancy develop greater competency through guided reflection.
What was the problem?
Grobot has been working with the Kent IT Consultancy, who are a student staffed IT Consultancy based at the University of Kent in the UK. Final year CompSci students work in the consultancy as part of an academic module and learn consultancy skills by delivering services to local businesses. Student Consultants earn academic credit by creating reflective reports based on their work in the consultancy, to show that they have met a series of learning outcomes.
The KITC management team were worried that the students weren’t taking the time to reflect on their experiences soon after they happened, often waiting until just before the report deadline, which could be weeks later. They reasoned that this was partly due to their focus being on getting their consultancy projects done, but also that the students struggled to know how to reflect effectively on an experience. Where possible the management team help the students , guiding their reflection during 1:1 conversations, but this doesn’t scale well, and isn’t doesn’t allow a consistent experience.
What was the solution?
The KITC uses Slack as its primary communications channel. Consultants use it to plan and execute their projects, as well as provide updates and reports to the management team. We wanted to create a way that consultants could reflect on their work, in a way that kept them in their workflow, and that they could do whenever they liked, so we built a “personal development coach-bot” that lives in Slack.
The chatbot is launched by invoking the “/log” command inside a direct message channel with Andre:
As well as being able to select what project a reflection relates to, the consultant has the option to choose one of three types of reflections.
Lightning, as the name suggests is super quick, but powerful. It asks the consultant to answer just two quick questions which are designed to help them quickly review what happened and to decide on actions to take in the future.
Quick Reflection is slightly more in-depth, and requires the consultant to spend some time doing some quantitative and qualitative analysis of their experience.
The last option takes the consultant on a complete loop of the Gibbs Reflective Cycle — the gold standard for academic reflection.
As you can see in the gif above, depending on the consultant’s choice Andre then takes them through a dialogue, asking them a series of questions, allowing the consultant as much time as they want to respond.
At the end of the sequence Andre creates a Slack Post that replays the conversation. The consultant can then review, share or download these posts. This means that when it’s time to write-up their reflective reports at the end of a project, all their reflections are in one place.
What have we learnt?
The experience of the KITC’s management team was that having a human-to-human conversation with the same kinds of reflective questions was hugely beneficial to the student consultants’ development. We also knew that these conversations didn’t scale and it was rare that conversations like this took place, so the opportunity to automate one side of the conversation was huge. Our concern however was whether the consultants would open-up in the same way and derive the same level of value from talking to a bot about their experience? We have gone live to a small group of early adopters, and so far the feedback is encouraging:
Andre has allowed me to reflect on my weekly work for projects in the KITC in a more effective and efficient manner. Not only does Andre guide me through the reflective process, it also helpfully stores past reflections so I can look back on them when writing-up my university assignments.
Tom Matthews — KITC Student Consultant
Aside from being a novel use case, we think that Andre is a great example of a simple (no complex AI or NLP here) internally focussed chatbot, that does one thing well and allows an important internal process in the KITC to scale better.