university-of-glasgow-chatbots

Saving 9,000 Staff Hours With AI Powered Student Support

Snapshot

The Numbers

With Gecko, the University of Glasgow has handled nearly 100,000 student conversations, saved around 9,000 staff hours, and increased chatbot success rates from around 93% to over 95%. Average response times are now under six seconds, with student satisfaction scores rising by roughly 17%.

100,000

unique conversations

~9,000

staff hours saved

<6 seconds

response time

From Challenge to Impact

Challenge

Original chatbots had a 93% message success rate but had gaps, questions remained unanswered, and staff wished to improve response speed, broaden coverage, and scale up without increasing manual workload.

Action

They integrated ChatGPT with their existing bots so that Gilbert handles tricky questions via a hybrid model and the Short Courses Bot takes fuller control in its niche, also feeding both bots with real time data from the website, and enabling suggested answers for human agents in live chat.

Impact

Success rates improved from ~93% to over 95%. Over 1,300 newly stumped questions are now answered. Staff saved ~9,000 hours. Multilingual support was added. Average response times dropped under 6 seconds. Student feedback (“great” or “amazing”) rose by ~17%.

The Full Story

University of Glasgow (UoG) built on their earlier success with their main chatbot “Gilbert” (launched Nov 2020) and a Short Courses Bot (late 2021) by integrating them with ChatGPT. The idea was to supplement their bots’ capabilities. Gilbert continued to handle broad inquiries but was enhanced to automatically pick up harder questions via ChatGPT. The Short Courses Bot, with more constrained scope, was given more autonomous control using ChatGPT. Both bots received real time data from the Glasgow website to stay up to date.

Within twelve days of integrating ChatGPT, they saw measurable improvements. Success rates rose above 95%. Over 1,300 previously unanswered questions could now be handled. Students got responses in under six seconds on average, and feedback from those involved rose significantly. They also enabled multilingual support (including English, Hindi, Arabic, Mandarin), widened reach and improved satisfaction.

Glasgow also changed their enquiry infrastructure. They removed their phone based call center during peak periods, replacing it with the bot plus live chat plus form plus enquiry modules. This reallocation of effort allowed staff to focus on quality rather than volume, and students benefited from almost instantaneous answers rather than delayed emails or calls.

Their experience shows how rapid deployment (under a minute to set up the ChatGPT component), smart use of hybrid human and AI approaches, and feeding AI with real time data can yield big gains in satisfaction, coverage, and efficiency, without proportionally increasing staffing.

Client

University of Glasgow

Type

University

Enrollment Size

20,000 and above

State / County

Scotland

Country

United Kingdom

How Glasgow used Gecko to boost student support