Deploying AI Assistant Hanna to 500 Staff: Practical Lessons

21/10/2026

The smallest technical glitches often cause the most significant bottlenecks in AI projects.

The challenge facing a major Vietnamese retail corporation was clear: they needed an office AI assistant capable of handling massive daily volumes of emails, meetings, and documents. The goal was to reduce the workload for their operations team. We, AIVISION, were brought in to consult and deploy the AIV Office system, specifically focusing on the Hanna AI assistant. This was a standout project because it was not just about installing software; it was about changing the work habits of hundreds of employees.

The Initial Challenge: Bottlenecks in Office Workflows

Before starting the work, we sat down with the operations department to listen. The issue wasn't a lack of tools. They already had email, calendars, and project management software. The problem lay in fragmentation and waiting times. An employee could spend nearly an hour just locating an old document scattered across multiple folders and inboxes. Drafting partner response emails or compiling reports from complex spreadsheets took far longer than necessary. They needed a solution to connect these pieces, but they didn't want to add another cumbersome application to the list of tools they open every morning.

Real-World Conditions and Human Challenges

This is often the most difficult and underestimated aspect of office AI projects. Employees are accustomed to habitual workflows. When introduced to a new office AI assistant, the initial reaction is often skepticism or resistance to change. Many were concerned that using Hanna would expose sensitive information or lead to job losses. In reality, psychological barriers are often greater than technical ones. We realized that simply providing a powerful tool without addressing trust and convenience would cause the project to fail during the pilot phase. Some employees even intentionally ignored AI suggestions to revert to familiar manual methods, even when they knew those methods were slower.

Implementation Strategy: Deep Integration into AIV Office

Instead of forcing employees to open a new tab or a separate application to interact with the AI, we decided to integrate Hanna directly into the familiar workflows within the AIV Office ecosystem. When a user is drafting an email in Mail, they can ask Hanna to summarize content or suggest a reply right within the editor. When viewing a spreadsheet in Office, they can ask Hanna to analyze data trends and receive immediate results. This approach reduces friction in the user experience. We did not deploy all features at once. Instead, we started with small, repetitive, and low-risk tasks, such as summarizing long documents or searching for files. Only when employees saw clear value and trusted Hanna's accuracy did we expand to more complex tasks, such as preparing meeting agendas or conducting deeper project analyses. This process required patience and short, hands-on training sessions rather than lengthy theoretical lectures.

Measurable Outcomes and What We Would Do Differently

After a few months of operation, we did not measure success by mechanically counting saved hours, but focused on qualitative metrics and satisfaction. Employees began proactively asking Hanna before searching manually. The rate of meetings delayed due to missing information dropped significantly. More importantly, the work atmosphere became lighter as tedious administrative tasks were handled efficiently. However, there were areas for improvement. Some user requests were too vague, requiring Hanna to ask clarifying questions, which caused delays. This highlights that the user's ability to ask questions is just as important as the model's quality. If we were to do it again, we would dedicate more time to onboarding, creating standard question templates, and training AI interaction skills from the start. This helps shorten the adaptation period and improves the accuracy of outputs. This project demonstrates that successful office AI deployment depends not on having the smartest algorithm, but on deeply understanding people and their daily work habits. AIV Office with Hanna is the tool, but the shift in mindset is the foundation.

A small step you can take this week: Select a group of 5-10 employees with the highest office workload, place them in the AIV Office environment, and ask them to use Hanna to handle a simple task, such as summarizing a long report or drafting a response email. Record their feedback on ease of use and accuracy. This is the fastest way to determine if an office AI assistant truly fits your company's work culture.

AIVISION helps enterprises turn AI into working systems. Explore our enterprise AI solutions, read more on the AIVISION blog, or talk to our team about your own use case.