AI Agents Are Moving From Chat to Doing
For many Malaysian SME owners, artificial intelligence has so far meant asking a chatbot to draft a Facebook caption, summarise a document or suggest ideas. The next stage is more practical—and more demanding. AI agents are being designed to connect to your business tools, interpret information and complete multi-step work on your behalf.
That could mean preparing a weekly sales report, checking customer messages, organising information from email, updating a planning document or turning a conversation into a task list. It could also mean giving software access to sensitive company information. The question is no longer simply whether AI can produce a useful answer. The real question is whether you are comfortable allowing it to act inside your business.
A recent TechCrunch report describes OpenAI’s effort to make AI agents useful beyond software development, with access to workplace applications and digital workflows. For a Malaysian SME with a small team, this trend matters because automation may soon be less about buying another standalone application and more about assigning an intelligent digital assistant to repetitive coordination work.
What Happened
OpenAI is developing an agent-focused version of its workplace software intended to help non-engineering workers complete complicated tasks across the applications they already use. According to TechCrunch, the product is designed around the idea that an AI system should do more than answer questions: it should carry out a project over several steps, using workplace information and connected tools.
The report says OpenAI’s internal teams use agentic software for activities such as weekly metrics reporting, spreadsheet-based planning, dashboards and data visualisation. It also describes examples involving investment research, internal communications and turning workplace conversations into charts. These examples show where agents are initially most useful: structured work that involves collecting information, organising it and producing a repeatable output.
However, adoption outside technology teams remains a challenge. An OpenAI-backed study cited by TechCrunch found that 98% of OpenAI employees were using its coding agent in June, compared with 17% of organisational subscribers and less than 1% of individual subscribers. The figures indicate a familiar gap: technology companies may understand how to use advanced tools, while ordinary businesses still need simple interfaces, clear guidance and confidence that the system will not make dangerous mistakes.
“How much control are you willing to give an LLM over your digital life?” That question, raised in the TechCrunch report, should become a management question for every business owner—not only a technology question.
Why This Matters for Malaysian SMEs
Most Malaysian SMEs do not have separate departments for administration, customer service, marketing, finance and operations. One person may handle sales follow-ups in the morning, supplier coordination after lunch and social media updates in the evening. An AI agent could help connect these activities instead of treating each task as an isolated prompt.
Consider a local food manufacturer receiving orders through WhatsApp, email and an online form. An agent could potentially gather the requests, identify missing delivery details, create a draft order summary and prepare a follow-up list for the team. It should not automatically confirm every order or change records without approval, but it could reduce the time spent copying information between channels.
A services firm could use an agent to review incoming enquiries, classify them by service type, prepare draft replies in Bahasa Malaysia or English and highlight urgent leads. A retailer could ask an agent to combine weekly sales information with stock records and identify products that require attention. A small construction or maintenance company could turn site updates into a structured progress report for the customer.
The opportunity is especially relevant when your staff already work across familiar tools such as spreadsheets, email, cloud storage, messaging platforms and accounting systems. The value does not come from adding another dashboard. It comes from reducing the handoffs between systems and people.
| Business area | Possible agent assistance | Human approval needed |
|---|---|---|
| Customer service | Sort enquiries and draft replies | Complaints, refunds and sensitive cases |
| Sales | Summarise leads and prepare follow-ups | Quotes, discounts and commitments |
| Operations | Combine updates into daily reports | Schedule changes and resource decisions |
| Finance administration | Organise documents and flag missing details | Payments, approvals and statutory submissions |
| Marketing | Repurpose approved content for channels | Public claims, brand tone and final publishing |
The Bigger Picture
AI agents are part of a broader shift from software that waits for instructions to software that can pursue an objective. Instead of opening five applications and moving information between them manually, you may eventually describe the outcome you want and let the agent coordinate the steps.
That does not mean you should hand over your entire business immediately. The TechCrunch report highlights the risks of connecting agents to private messages, workplace applications and personal information. An agent may find relevant material in the wrong place, misunderstand permission boundaries or include confidential information in a document that was meant for a wider audience.
For Malaysian SMEs, governance should begin before adoption. Create separate access levels for owners, managers and staff. Keep customer, employee and financial information in approved systems. Decide which tasks are read-only, which can create drafts and which—if any—can make changes automatically. Review activity logs regularly and remove access when a staff member leaves.
You should also define what “good” looks like. Start with a task that is frequent, time-consuming and easy to check. A weekly management summary is usually safer than allowing an agent to approve supplier payments. A draft customer response is safer than an unsupervised promise about delivery dates. Measure the result by checking accuracy, time saved and the number of corrections required.
How to Prepare Your Business
- Map repetitive workflows. List tasks involving copying, sorting, summarising or preparing standard documents.
- Choose one low-risk pilot. Begin with internal reporting, meeting summaries or draft follow-ups.
- Clean up your data. Clear file names, consistent customer records and defined folders make automation more reliable.
- Set approval rules. Require a person to review external messages, financial actions and sensitive documents.
- Limit permissions. Give the agent access only to the applications and folders required for its assigned task.
- Train your team. Staff should know what the agent can do, what it cannot verify and when to escalate an issue.
The companies that benefit most will not necessarily be the ones with the biggest technology budgets. They will be the ones that understand their processes clearly and introduce automation with sensible controls. AI agents may eventually become as ordinary as shared spreadsheets or online accounting systems, but trust will determine how quickly they enter your daily operations.
For now, you do not need an agent for everything. Choose one business process where your team loses time every week, keep a human in charge of the final decision and test whether the technology genuinely improves the workflow. That is a practical first step towards a more automated SME.
Source: TechCrunch, “OpenAI is building AI agents for everything.
Ready to Streamline Your Operations?
Technology moves fast. Your operations should keep up. AutoRunBiz builds AI systems that run your daily workflows — from WhatsApp order capture to accounting. Book a free 15-min ops audit →