Meet us @ Booth #101, EDUCAUSE . October 27–30, 2025 | Nashville
How Broward College replaced a failed legacy bot, deployed Chat Al and an Academic Advising
Voice Agent, and cut after-hours agent workload by 45% without cutting service.
Broward College OneStop is among the most complex student services environments in community college education. More than 81,300 calls per year. 42% of them academic advising. Peak months that hit 12,550 calls and overwhelm every capacity model.
The institution knew it had a volume problem. What it discovered, through a failed first attempt at automation is that it also had a quality problem.
Broward College OneStop handles an enormous volume. Nearly half of every call is an advising question FAFSA status, registration blocks, hold removal, degree requirements. Peak months like August hit 12,550 calls alone. Agents couldn’t absorb the volume without queues that stretched 15-20 minutes or more.
The existing Dialogflow voice bot handled 8x human agent volume but 48% of those interactions ended as unsuccessful handoffs. Students who couldn’t get answers from the bot called back for a human. The tool designed to reduce calls was generating them, and every failed interaction eroded trust in self-service.
With 27 agents and no single source of truth, students got different answers depending on who picked up. Knowledge base updates required manual retraining that didn’t reach frontline staff instantly. The gap between what the institution knew and what students were told widened every time a policy changed
Advising questions don’t stop at 5 PM. Students registering late at night, checking financial aid status over the weekend, or navigating a hold before an enrollment deadline had no self-service path. Full after-hours staffing was cost-prohibitive Voicemail was the Iternative. Neither served the student
The legacy Dialogflow bot was eventually discontinued not because automation was the wrong answer, but because that particular tool was the wrong tool. The question Broward brought to BlackBeltHelp was not whether to automate, but how to do it in a way that actually worked for students.
BlackBeltHelp approached Broward College’s challenge as a two-channel problem: students asking questions in chat needed a smarter agent, and students calling for advising after hours needed someone to pick up.
The difference between the legacy bot and BlackBeltHelp is not the channel it’s the understanding. Students don’t have to phrase questions correctly. They don’t get routed to a handoff when the intent doesn’t match. They get an answer.
The Chat Al agent draws answers from the institution’s own knowledge base and syncs automatically when content changes no manual retraining, no lag between policy update and agent knowledge. The Academic Advising Voice Agent operates after hours, handling FAFSA verification, registration blocks, hold removal, QLess queue management, transcript requests, and major changes. When a question is too complex, it escalates with context to a Live Human Agent.
The voice agent went through a deliberate iteration cycle: customer testing in January 2026 revealed that knowledge base coverage needed to expand and the agent needed to probe more deeply for context. Those improvements were made. The resolve rate went from 14% at launch to 45.5% by March 2026. That improvement is the model test, learn, improve, and expand.
Broward College’s path included a failed first deployment, which makes its current results more meaningful, not less. The institution learned what didn’t work and built something that does.
With Chat Al and the Academic Advising Voice Agent live, Broward College has measurably shifted how
student services are delivered more capacity, more consistency, and 24/7 availability that didn’t exist before.
Students waited 15–20+ minutes for simple
advising questions peak months hit 12,550 calls
with queues overflowing
27 agents with varying knowledge gave
inconsistent answers students got different
information depending on who picked up
48% of calls through the legacy Dialogflow bot
ended as unsuccessful handoffs students gave
up or called back
After-hours advising was either fully staffed at full
cost, or unavailable students got voicemail
Knowledge base updates required manual agent
training changes didn't reach frontline staff
instantly
The Chat Al and Voice Al agents pick up instantly.
Zero queue time. Students get answers the
moment they ask.
Every answer comes from the same knowledge
base. Consistent, accurate information every time,
across every channel.
The Academic Advising Voice Agent handles
after-hours calls at a 45.5% resolution rate, up
from 14% at launch.
50% of after-hours calls are now handled by the
Voice Alagent. Agents focus on complex cases
during business hours
The Chat Al agent syncs from the knowledge base
automatically. Updates go live immediately, with no
retraining required.
"We had tried automation before and it made things worse. What changed with BlackBeltHelp is that students actually get answers and the ones who need a human reach one faster because the Al is handling everything else."
- Broward College - OneStop Student Services
Broward College's experience is a case study in what it takes to make student services automation actually
work. The first attempt a rigid, intent-based system failed because it was built around what the
technology could do, not what students needed. The second attempt succeeded because it started with
the student experience and worked backward.
The 45.5% resolve rate on the Academic Advising Voice Agent isn't a ceiling it's a starting point. The gap
between 14% at customer testing and 45.5% two months later is a direct result of iteration: better knowledge
base coverage, deeper conversational probing, faster syncing from updated content. The model improves as
it learns.
For a community college serving a high-need student population, the stakes are particularly high. A student
who can't get an answer about a registration hold on a Sunday night doesn'tjust experience inconvenience
they may not enroll for the following semester. 40% of stop-outs cite financial confusion or administrative
friction as a factor. Broward is addressing those friction points in real time.
With Chat Al and an after-hours Academic Advising Voice Agent live, Broward has done what most institutions are still planning: deployed Al that students actually use, iterate on, and benefit from. In our experience, institutions at this stage of the journey ask the same three questions about what comes next.
The Academic Advising Voice Agent is live after-hours and proving its resolve rate. The natural next step for institutions at this stage is extending that coverage to business hours absorbing peak daytime volume and letting agents focus on the cases that require direct human engagement.
The Chat Al agent currently answers general questions. Once connected to Banner, it answers student-specific questions their registration hold, their aid disbursement status, their degree audit. The same platform, now personalized to every student who interacts with it.
Replace the complex IVR too many menu levels, wrong department routing, long hold times with natural language intent routing. Students say what they need and reach the right place on the first attempt. Every call answered, every caller routed correctly, around the clock.
BlackBeltHelp is the #1 AI-powered Helpdesk platform purpose-built for higher education. It combines Generative AI, trained agents, and integrations with systems like Ellucian™ Banner, Canvas, Slate, and Microsoft Entra to answer IT, student services, and admissions questions. Trusted by 130+ institutions including Columbia University, Cornell University, and the California Community Colleges System, BlackBeltHelp manages over 8 million helpdesk interactions annually, ensuring 24x7 support so campuses stay connected and students stay supported.
