Good morning, everyone! The Knowledge-Ba...

Audio by Paper2Audio

Good morning, everyone! The Knowledge-Based Chatbot Innovation Project is brought to you by Beverly Mendola, Julius Kim Cruz, Michael Santos, Michelle Puda, and Mirasol Reyes of the Data Collection and Parsing team.

Our team is primarily responsible for working with internal and external stakeholders to successfully complete implementation orders, and for parsing set of reports, which are then translated into a macro-based workbook called the Data Source File, or D.S.F.

Before we introduce the project, let's start with the problem we set out to solve. Whenever someone on the team has a process-related question, the only option is to go through lengthy documentation and manuals, which takes time. Different people often interpret the same documentation differently, leading to rework, missed deadlines, and inconsistent output across the team.

To solve this, we built a knowledge-based chatbot on Microsoft 365 Copilot technology, configured specifically as a knowledge-base assistant for the Data Collection and Parsing team.

The chatbot is configured according to our team's desired tone, behavior, and response preferences, so it feels consistent with how our team communicates. We provided it with our semantically condensed and compressed reference materials, including our manuals, work instructions, and process documentation, rewritten specifically for A.I interpretation. This matters because raw, long-form documentation isn't structured the way an A.I model reads best. Condensing it down to its core meaning helps the chatbot retrieve the right answer quickly and accurately, instead of pulling irrelevant or overly broad information from lengthy pages.

The chatbot guides employees through parsing operations, data collection activities, implementation workflows, quality assurance processes, and other approved knowledge areas. This is meant to be a fast reference tool. It doesn't replace guidance from Senior team members or Team Leads when deeper clarification is needed.

Now that we've covered what the chatbot does, let's look at how to access it. Once the chatbot is published, access is granted by sharing. From there, access can be given to specific users, groups, Microsoft Teams users, or the entire organization, if permitted by the administrator. Once shared, the chatbot appears directly in Copilot within Teams. Anyone in the organization with Microsoft 365 Copilot can open Copilot in Teams, select and choose our chatbot from the list of available agents.

With accessibility covered, let's talk about what this actually means for the team. The biggest benefit is speed. Instead of stopping work to search through documents or interrupt a teammate for an answer, users get a quick, reliable response grounded strictly in the approved reference material. This means less downtime for the person asking, and less disruption for the person who would've been asked.

Looking ahead, the next step is to keep reference materials continuously updated, so the chatbot's answers stay accurate as processes evolve. The team is also looking into expanding this beyond D.C.P, potentially developing a version for external users so the same fast, reliable support can extend past the team. Longer-term, the plan is to build in a feedback loop, so as users interact with the chatbot, the team can identify gaps in documentation and continuously improve both the chatbot and the source material behind it.

That will be all and thank you so much for your time!