ProServ Logistics – Custom RAG Chatbot

Deployed an internal AI agent trained on Google Drive docs, providing 24/7 instant answers and saving the HR/management team 10+ hours per week on repetitive questions

The Challenge

ProServ Logistics, a company with over 100 employees (many of them remote), struggled with knowledge management. All key documents—Standard Operating Procedures (SOPs), HR policies, safety manuals, and technical guides—were stored in hundreds of files (PDFs, .docx) on a shared Google Drive.

New employees spent weeks on onboarding, and existing staff constantly bombarded managers and HR with the same questions: “What is the procedure for X?”, “Where can I find form Y?”. Answering simple questions took up valuable hours, and finding the right document was frustrating and inefficient.

The Solution

The NexaroAI team was brought in to automate this knowledge bottleneck. Instead of building another complex portal, we implemented an intelligent, internal AI assistant (a RAG chatbot) directly into their existing workspace.

Our 3-step implementation process:

  1. Data Ingestion: We created a workflow that automatically monitors ProServ’s dedicated Google Drive folder. Every new or updated document is immediately fetched, split into smaller chunks, and processed.

  2. Creating the “Brain” (Vector Database): Each text chunk was converted into a vector (a numerical representation) and saved in a secure Pinecone Vector Store. This created the bot’s “brain,” containing only the company’s knowledge.

  3. Agent Deployment (AI Chatbot): We built a simple chat interface integrated with their company Slack. When an employee asks a question (e.g., “How many vacation days do I have?”), the workflow:

    • Captures the question.

    • Searches the Pinecone database to find only the document snippets relevant to the answer.

    • Passes the question and the found snippets to the Gemini AI model with the instruction: “Answer this question using only this provided information.”

The Results

The deployment of the internal AI Agent took less than two weeks and immediately transformed how the company operates.

  • 70% Reduction in Repetitive Questions: Managers and the HR department reported an immediate drop in queries about standard procedures.

  • 24/7 Availability: Employees (especially across different shifts) gained instant access to accurate information without waiting for a response.

  • Faster Onboarding: New hires can now “talk” to the company documentation, reducing the time needed for onboarding by an estimated 30%.

  • 100% Data Security: The RAG solution ensures the AI only answers based on the company’s private documents and does not “hallucinate,” and the data never leaves their secure environment.

The results in numbers

25h
Saved Per Week
24/7
Knowledge Access
% Less
Repetitive Questions
%
Faster Onboarding
<2 Months
Payback Period
CASE STUDY Q&A

Key Project Details

What specific AI technologies were used?

This solution was built on a RAG (Retrieval-Augmented Generation) architecture. We used n8n to connect Google Drive, a Supabase Vector Store for the ‘knowledge brain’, and Google’s Gemini AI model to generate the accurate, context-aware answers.

How long did this implementation take?

The initial proof-of-concept was ready in days. The full, production-ready solution, including indexing all documents and integrating with Slack, was designed and launched in under two weeks, fitting within our ‘Business’ package timeline.

Is this solution secure? Does the AI 'learn' from private data?

Yes, it’s highly secure. The RAG model ensures the AI only *reads* your private documents to find an answer; it does not *train* on them. All data stays within your secure environment (Google Drive and Supabase), and the AI cannot answer questions about anything outside that knowledge base.

What kind of maintenance does this system require?

Minimal. The n8n workflow automatically re-indexes any new or updated documents you add to the Google Drive folder. The monthly Maintenance plan covers all system monitoring, bug fixes, and performance optimization, making it a “set it and forget it” solution for the client.

Related Case Studies

Biznes ✪ Social Media

Transformed a 3-hour manual video workflow into a 3-minute automated task. We used n8n and AI to automatically generate video and publish it to 9 social platforms.