AI Feedback Is Moving From Experiment to Everyday Practice
If you run a Malaysian SME, you probably do not have time to coach every sales pitch, customer proposal, staff presentation, or management meeting. You may be the person approving campaigns, reviewing quotations, training new hires and answering operational questions—often in the same day.
That creates a familiar problem: your team needs regular feedback, but your schedule cannot support constant one-to-one coaching. Harvard Business School’s use of AI avatars in its eight-week Foundry entrepreneurship programme shows one possible direction. According to TechCrunch, participants can practise pitches and board meetings with AI versions of instructors, while live sessions with human instructors continue weekly.
TL;DR: AI avatars are not a replacement for your judgement or personal relationships. They are a practice layer that can help staff rehearse, receive immediate feedback and improve before an important customer or management conversation.
For a small business, the useful lesson is not whether an AI avatar looks perfectly human. The useful lesson is whether your team can practise more often, using feedback based on your company’s standards.
What This Means
An AI avatar is a digital representation of a person that can speak, listen and respond during a simulated conversation. In the Harvard example, the avatars are based on instructors and are used to provide feedback during exercises such as practice pitches and board meetings, as reported by TechCrunch.
Think of it as a rehearsal partner rather than an automated manager. Your employee could practise explaining a new service, respond to a difficult customer question or present a monthly business update. The AI could then assess whether the explanation was clear, whether important information was missing and whether the speaker answered the question directly.
The quality of the exercise depends on the instructions and information provided. If you give the system a vague role, it may produce generic comments. If you define your customer profile, service rules, brand tone and evaluation criteria, the feedback can become more relevant.
Human involvement still matters. The Harvard programme includes live instructor sessions alongside the avatar experience, according to TechCrunch. That combination is important for SMEs: AI can support frequent practice, while you or an experienced manager handles judgement, context and sensitive decisions.
The practical insight: Use AI to increase the number of practice opportunities, not to remove human responsibility from important decisions.
How This Applies to Malaysian SMEs
1. Sales teams can practise before meeting customers. If you sell software, industrial equipment, renovation services, training or professional services, new staff often struggle with questions about suitability, implementation and results. You can create a simulated customer who asks common questions in English, Bahasa Malaysia or Mandarin, depending on your market. The employee practises their response, receives feedback and tries again before speaking to a real prospect.
For example, a Klang Valley renovation company could ask staff to explain project timelines, material choices and change-request procedures. A distributor in Penang could simulate a buyer who is comparing suppliers and demanding faster delivery. The purpose is not to generate a perfect script. It is to help the employee explain clearly without making promises that your business cannot keep.
2. Customer service teams can rehearse difficult conversations. Malaysian SMEs often depend on a small number of people to handle complaints, delayed orders and refund requests. One poorly handled WhatsApp conversation can create unnecessary escalation. An AI practice assistant can play the role of an impatient customer, a confused first-time buyer or a client asking for an exception.
You can define the boundaries: what your staff may approve, when a supervisor must be involved and which phrases should be avoided. Staff can practise showing empathy, confirming the issue and offering the next step. The final decision should remain with an authorised person, but your team becomes more prepared for the conversation.
3. Managers can improve internal presentations. Many SME owners promote technically capable employees into supervisory roles without formal presentation training. These new managers may know the work but struggle to explain a late project, staffing issue or operational risk during a meeting. A simulated manager or board member can ask follow-up questions such as, “What caused the delay?”, “What will you do next?” and “How will you know the problem is fixed?”
This is especially useful before monthly reviews. The employee can practise presenting numbers, explaining exceptions and responding without becoming defensive. You can then spend your meeting discussing the real issue rather than teaching basic presentation structure.
4. Owners can turn company know-how into repeatable training. Your best salesperson or operations supervisor may carry years of practical knowledge in their head. If that person leaves, new staff may have to learn through trial and error. An AI role-play system can be guided by your standard operating procedures, approved answers and real customer scenarios.
Be careful with confidential information. Do not upload customer identification details, private contracts, passwords or sensitive financial records into a tool without reviewing its security and data-handling terms. Start with generalised examples and remove names, account numbers and other unnecessary details.
A Simple SME Implementation Plan
| Stage | What You Do | Suggested Measure |
|---|---|---|
| Week 1 | Choose one repeatable scenario, such as a sales pitch or complaint response. | List 5 common questions and 3 unacceptable responses. |
| Week 2 | Create a role-play using your approved process and tone. | Ask 2 staff members to complete 3 practice rounds. |
| Week 3 | Review the AI feedback with a manager. | Identify the 3 most common communication gaps. |
| Week 4 | Update the scenario and add a second employee group. | Compare improvement across 2 or more practice attempts. |
The figures in this plan are operating targets rather than industry benchmarks. Adjust them to your team size and workload. A five-person business may need only one scenario, while a 50-person company may create separate exercises for sales, service and supervisors.
Practical Takeaways
- Start with one high-frequency situation: choose a conversation your staff handle every week.
- Define success clearly: list the information that must be included, the approvals required and the tone expected.
- Use real Malaysian contexts: include WhatsApp enquiries, bilingual conversations, delivery delays and follow-up expectations where relevant.
- Keep a human reviewer: a manager should check whether the feedback matches your actual business policy.
- Protect business information: remove personal data and confidential details before using any external AI tool.
- Measure behaviour, not novelty: check whether staff explain more clearly, follow procedures and escalate issues appropriately.
- Do not copy a famous personality blindly: your team needs feedback aligned with your customers and operating reality.
Questions to Ask Before You Adopt It
Before selecting an AI avatar or role-play tool, ask who owns the recordings, whether conversations are used to improve the provider’s system and where data is stored. You should also check whether administrators can delete sessions and control access. These questions matter more than whether the avatar looks realistic.
Ask your employees for practical feedback too. Some people may prefer text-based practice, while others may benefit from speaking aloud. If the tool makes staff uncomfortable, provide a non-video option. The goal is better preparation, not a performance contest involving technology.
Finally, decide what the AI is not allowed to do. It should not approve discounts, promise delivery dates, interpret legal obligations or make employment decisions unless a qualified human reviews the result. Clear boundaries will make adoption safer and easier to manage.
The Bigger Picture
The longer-term trend is a shift from one-off training to continuous practice. Traditional training usually happens during a workshop, followed by limited follow-up. AI-supported role-play could let employees rehearse whenever a new product, policy or customer segment is introduced.
That does not mean every company needs an AI avatar of its founder. In many cases, a simple text or voice simulator will be enough. The important change is that coaching can become part of the workflow instead of something reserved for senior staff or occasional training days.
For Malaysian SME owners, the strongest approach is to combine automation with your own judgement. Document your best practices, turn common situations into exercises and let staff practise before mistakes reach customers. Then use human reviews to improve the scenarios over time.
Harvard’s example, reported by TechCrunch, points to a practical principle: people may accept AI assistance when it helps them prepare for a real task and when human guidance remains part of the process. Your SME can apply the same principle without copying the exact format. Start small, protect your information and focus on the conversations that matter most to your business.
Ready to Streamline Your Operations?
Your business should run itself. AutoRunBiz deploys AI agents to automate your daily operations — WhatsApp orders, invoicing, customer follow-ups, and accounting. Book a free 15-min ops audit to see where automation fits your business →