When Students Ask the Question, They Should Get the Answer.
How Columbia University reimagined student and staff support for IT and Procurement and freed their teams to focus on work that actually needs a human.
How Columbia University reimagined student and staff support for IT and Procurement and freed their teams to focus on work that actually needs a human.
Columbia University’s IT department and Procurement team were both using legacy chatbots that had the same fundamental flaw: they could only answer questions they’d been specifically programmed to expect.
That sounds like a technology limitation. In practice, it was a student experience problem and a staff burden problem.
The legacy chatbot required exact phrasing. A slight variation in how a student asked a question even about something it could technically answer produced no result. Instead of reducing calls and emails to staff, the chatbot added frustration.
Password resets were repetitive, pattern-based work:
Every policy update, every new process, every change to IT or procurement procedures had to be manually translated into new chatbot responses. Keeping the system current was a job in itself and when it fell behind, users lost confidence in it.
Building responses, updating answers, and testing new intents required dedicated staff time. The tool meant to reduce the support burden was creating one. Meanwhile, students and faculty were still calling and emailing for answers the system should have provided.
IT support and procurement have different language, different workflows, and different user needs. Serving both required running and maintaining two separate systems doubling the effort without creating any shared efficiency.
The result was a support model that promised self – service but delivered frustration.
Students called instead of self – serving. Staff maintained chatbot responses instead of advising. And as the university grew and policies changed, the gap between what the system knew and what students needed kept widening.
Columbia partnered with BlackBeltHelp to reimagine how support works not by building a better version of the old tool, but by replacing the underlying approach entirely.
Instead of asking students and staff to learn how to talk to the support system, BlackBeltHelp built a support system that understands how they already talk.
BlackBeltHelp’s Digital Agents draw answers from the institution’s own documentation knowledge bases, websites, policy documents and keep those answers current automatically. There are no response lists to build, no configurations to update, and no intent libraries to maintain.
Crucially, BlackBeltHelp isn’t just a chatbot. When a question is too complex for self – service, Live Human Agents higher – ed trained and available around the clock step in. Students never hit a dead end. Staff don’t spend their day on questions technology can handle.
This is what modernized student support looks like: intelligent enough to answer automatically, connected enough to stay current, and human enough to know when to escalate.
Columbia’s modernization followed a deliberate path beginning with a legacy chatbot deployment and advancing, over fourteen months, to a fully live, self – maintaining support model across two departments.
Within weeks of launch, both Digital Agents were actively resolving student and staff questions without a single response list, and without staff intervention on routine queries.
Students had to phrase questions in exactly the right way anything else got no response
Every new question type required the IT team to manually build a new response from scratch
As policies changed, answers became outdated. Users stopped trusting the chatbot and called staff instead.
IT and Procurement each maintained separate chatbots, with separate maintenance burdens
Email support requests landed in staff inboxes and had to be triaged and answered manually
Students ask naturally, in their own words. The Digital Agent understands them regardless of how they phrase it.
The Digital Agent draws answers from the institution's own documentation automatically no manual configuration needed.
Answers stay current because they come from live documentation not a static list built months ago.
A single BlackBeltHelp platform serves both departments. One system, two Digital Agents, shared infrastructure.
An email - based Digital Agent is built and ready to auto - respond and route requests deployment pending.
"We went from spending staff hours keeping the chatbot current to spending none. The support agent answers from our own documentation accurately, without us having to touch it."
Columbia University - CUIT
What happened at Columbia isn't a technology story. It's a student experience story and a staff capacity story.
When support works the way students actually expect it to, the downstream effects are real: fewer calls to the help desk, fewer emails that go unanswered, fewer students who disengage because they couldn't get a basic question resolved. Staff who were spending time on repetitive queries are freed to do advising, not answering.
This is the promise of student support modernization not deploying technology for its own sake, but building a support model that scales without burning out the people running it.
Columbia's experience also demonstrates a key principle in how BlackBeltHelp approaches deployment: you don't have to change everything at once. The Digital Agents went live in Procurement first, then IT. Each deployment builds confidence, surfaces what's working, and creates the foundation for the next step.
With digital chat Al agents live across Al and Procurement, Columbia has done what most institutions haven't yet: proven that modernized support works at scale, across departments, without ongoing staff maintenance. In our experience, institutions at this stage of the journey tend to ask the same three questions about what comes next.
For most campuses, students still pick up the phone especially for urgent issues like account lockouts and password resets. Voice Al handles those calls automatically, 24/7, without hold time or staff intervention. It's the same knowledge base that powers the digital chat Al agent, now accessible by voice.
The most powerful expansion isn't adding more departments it's making answers personal. When a Chat Al agent is connected to your student information system, it stops answering general questions and starts answering the right question for that specific student: their registration hold, their financial aid status, their next steр.
Institutions that modernize IT support quickly ask: where else are our staff fielding questions students could answer themselves? Enrollment, Financial Aid, and the Registrar are the most common next step same platform, same approach, new knowledge bases built around each department's content.