I’m in the process of collating material on some of the things I’ve worked on that I can share here. For now, here’s a sample (an MVP portfolio, if you like) with a short description of each initiative.
Beth: a chatbot for reflection
Did you know that the first ever chatbot was a pretend-therapist?
By mirroring back parts of what users said as questions, Eliza fooled some users into thinking it was really interested and caring. But is “fooled” even the right word?
Eliza’s creator, Joseph Weizenbaum, was drawing on the techniques of Rogerian psychotherapy. When I read about this, I thought: why stop at mirroring? What if we took the techniques documented in frameworks like Julian Edge’s Cooperative Development and used them as strategies for a chatbot to help users reflect on their personal development?
And so I built Beth. This was part of my MSc research dissertation. While it’s not been Turing-tested, I did trial it with several participants, and adapted a version of Eliza as a control.
Headless content rapid prototyper
In my first Content Team job, at the University of Reading, we had a painfully slow CMS.
One of the problems with slow response times is that they make it hard to test and learn. With a slow system, there can be an incentive to collect all the requirements from stakeholders up front and then go away and spend time building the thing. By the time the build is finished, if anything’s wrong, it’s too late or can take too long to make the required changes.
So I signed us up to a tool called GatherContent and built a prototyper on top of its API. The prototyper was hosted on a Heroku account I set up, and I created basic content models within GatherContent that mapped to pre-designed components on the server to enable quick and easy page builds.
Without really knowing it at the time, I’d effectively built a simple headless CMS for prototyping – this enabled us to iterate on content quickly and get feedback from stakeholders each sprint. After an average of four sprints, we’d feel confident that we knew what we needed to deliver and we then built the sites in the Content Management System for real.
After I successfully managed the migration of the SSE website to our new headless Content Management System, I wanted a way for us to maintain quality across the site.
So I wrote some scripts that would check the site for the use of certain words, or links that went to external sites but did not open in new tabs, or images that were missing alt attributes for accessibility.
I showed these scripts to some software engineers I had worked with on the migration and we got approval to develop a dashboard that would showcase some of these things so that anybody could see at a glance where updates needed making.
I created a requirements template and asked one of my team to use it to gather opinion on pain points or inconsistencies (and crucially inconsistencies that were difficult to spot and update normally) so that we could then prioritise those needs for the development backlog.
The resultant dashboard had widgets for checking when versions of content across environments were out of sync (and whether they were in progress or not, so we’d know if this was a problem), where images were being used across the site and if there were device-specific versions missing, and how our Markdown-formatted links held up.
I hope to include summaries of these initiatives soon too.
- Intent poker
- Patching with atomic diffs
- Mapping out Content Design maturity
- Collaborating on a UX competency matrix