Why this AI shift matters to your Malaysian SME
AI is no longer limited to answering questions, drafting captions or summarising documents. The latest enterprise research from OpenAI shows a shift towards AI systems that can connect to business information, use tools and complete multi-step work for human review. [Source: OpenAI, “From assistance to execution”]
For you as a Malaysian SME owner, this matters because your business probably does not have separate teams for every task. The same people may handle sales follow-ups, quotations, supplier coordination, recruitment, marketing and customer service. An AI agent that helps complete repeatable workflows could give your team more operating capacity without forcing everyone to work longer hours.
The important point is not to chase the newest AI label. It is to identify where work is repetitive, information-heavy and easy to check. Examples include preparing a draft quotation from a standard price list, turning meeting notes into follow-up tasks, comparing supplier responses, or preparing a first draft of a monthly sales report.
What Happened
OpenAI published two studies examining how organisations and employees are using AI. The first, Enterprise Signals, focuses on agentic AI adoption among OpenAI’s enterprise customers. The companion working paper examines how usage spreads across companies, job functions and seniority levels. [Source: OpenAI, “From assistance to execution”]
The reports describe enterprise AI moving from assistance to execution. An assistant may explain how to prepare a presentation, while an agent may gather information from approved sources, create a draft and prepare related files for review. This distinction is significant because the AI is participating in a workflow rather than producing a one-off answer. [Source: OpenAI, “From assistance to execution”]
Several figures show how quickly this transition is developing. As of June 2026, Codex produced 64% of combined Codex and ChatGPT output tokens among OpenAI enterprise customers. OpenAI also reported that frontier firms—the top 10% of organisations by monthly AI usage—generated 8.3 times as many output tokens per active user as typical firms, compared with 2.6 times in January. [Source: OpenAI, “From assistance to execution”]
| Finding | What it suggests for your business |
|---|---|
| Codex produced 64% of combined enterprise output tokens in June 2026 | AI use is shifting towards longer, delegated tasks |
| Frontier firms generated 8.3 times more output tokens per active user than typical firms | Depth of use may matter more than simply having access to AI |
| Weekly Codex users grew 41 times in sales and 26 times in marketing since February | Agentic workflows are spreading beyond software teams |
| Early-career employees sent 13 more AI messages weekly than executives six months after adoption | Younger or newer staff may already have useful AI habits to share |
Why This Matters for Malaysian SMEs
Many Malaysian SMEs operate with fragmented information. Customer details may sit in WhatsApp conversations, quotations in spreadsheets, product information in shared folders and delivery updates in email. When a staff member needs to respond to a customer, they may spend more time searching for the right information than preparing the response.
A properly controlled AI workflow can help bring these pieces together. For example, your sales coordinator could ask an approved AI tool to review a customer’s previous enquiries, identify the relevant product information and prepare a draft reply. Your team member still checks the facts, confirms availability and sends the message, but the initial preparation becomes faster and more consistent.
Consider a Malaysian distributor serving customers in Kuala Lumpur, Penang and Johor. A simple AI workflow could classify incoming enquiries, identify whether the customer needs a quotation or technical answer, retrieve the correct product specifications and create a draft for the sales team. If your business serves customers in Bahasa Malaysia, English and Mandarin, you could also use AI to prepare language-specific drafts while keeping a staff member responsible for tone and accuracy.
In professional services, the opportunity is equally practical. An accounting, consultancy or design firm could use AI to turn meeting notes into action lists, prepare a document checklist and flag missing information before a human reviews the file. A recruitment agency could summarise candidate information against an approved job description, while a marketing team could convert one approved campaign idea into platform-specific drafts.
These use cases should not be treated as automatic decision-makers. They are better viewed as structured assistants that can carry out preparation steps under clear instructions. Personal data, client information and confidential documents should only be handled through tools and processes that your business has reviewed.
How to start moving from assistance to execution
You do not need to automate your entire company. Start with one workflow that happens frequently and has a clear review point.
- Choose a repeatable task. Look for work that follows a recognisable sequence, such as enquiry handling, quotation preparation, invoice follow-up or weekly reporting.
- Document the current process. Write down the information required, the steps taken and the person who approves the final result.
- Define the AI’s boundaries. State what it may read, what it may prepare and what it must never send or change without approval.
- Connect only approved information. Begin with a clean folder, knowledge base or selected business system rather than granting broad access.
- Measure useful outcomes. Track turnaround time, corrections required and missed follow-ups. Do not measure success only by the number of AI prompts.
- Improve the workflow with staff feedback. The employees using the process daily will identify unclear instructions and exceptions faster than management alone.
“Connect agents to the context and tools needed to complete valuable work; establish clear permissions, review, and governance; and help employees turn effective individual workflows into shared ways of working.”
— OpenAI, “From assistance to execution”
The Bigger Picture
The widening gap between frontier firms and typical firms is a warning against treating AI adoption as a software subscription alone. OpenAI reported that frontier firms used Plugins weekly at a higher rate than typical firms: 21% compared with 9%. Skills usage was also higher among frontier firms, at 19% compared with 3%. [Source: OpenAI, “From assistance to execution”]
For an SME, the lesson is straightforward: access to an AI tool does not automatically create better operations. Your advantage comes from turning good individual experiments into documented team workflows. If one staff member has developed an effective way to prepare customer replies, capture the steps, review the risks and make the method available to the rest of the team.
The research also found that agentic AI is spreading across knowledge-work functions. Since February, weekly active enterprise Codex users grew 108 times in legal, 41 times in sales, 41 times in recruiting and 26 times in marketing, compared with five times in engineering. [Source: OpenAI, “From assistance to execution”] This is relevant to Malaysian SMEs because the first useful application may not be software development. It may be sales administration, recruitment, marketing production or customer support.
OpenAI also reported that early-career employees used AI more frequently than executives, with early-career workers sending 13 more messages per week six months after adoption. [Source: OpenAI, “From assistance to execution”] You can use this insight by asking newer team members which tasks they have improved with AI, then reviewing those methods for accuracy, privacy and consistency.
The practical path for your business is controlled experimentation. Select one workflow, connect the minimum necessary information, require human approval and record what improves. As your team learns, expand from isolated prompts to repeatable processes that support sales, operations and customer service.
AI agents will not remove the need for business judgement. They may, however, reduce the time your team spends moving information between systems and preparing routine work. For a Malaysian SME, that can mean faster responses, clearer handovers and more time for the decisions that require your experience.
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 →