Why a Scientific AI Breakthrough Matters to Your Business
A researcher’s use of ChatGPT and Codex to search for new antimicrobial molecules may sound far removed from running a Malaysian SME. It is not. The important lesson is how AI helps a small, cross-disciplinary team search through enormous amounts of information, form better hypotheses, automate repetitive work, and decide what deserves human attention next.
In the reported research, AI models search biological sequence data for molecules that could potentially fight drug-resistant infections. The early search, which could take years, may be reduced to hours before laboratory testing begins, according to OpenAI’s account of the project. For you, the practical equivalent is using AI to move faster through customer records, sales reports, supplier information, documents, and operational questions—without pretending that an AI prediction is automatically correct.
What Happened
César de la Fuente and his lab study the genomes of living and extinct organisms to identify possible antimicrobial molecules. Their deep-learning systems look for patterns in biological sequences and prioritise candidates for further investigation. The work responds to a serious global challenge: bacterial antimicrobial resistance was associated with about five million deaths in 2021, with the annual toll projected to roughly double by 2050, according to the source article.
The lab also uses ChatGPT and Codex for brainstorming, code writing, data processing, dataset organisation, result analysis, terminology checks, and connecting ideas across biology, chemistry, computer science, and engineering. Researchers with strong biology knowledge can get help with programming, while programmers can work more effectively on biological questions. The team still relies on laboratory experiments to validate predictions, and de la Fuente emphasises the need to double-check AI output, as described in the original report.
This distinction is essential. AI does not replace the laboratory, clinical testing, regulatory review, or scientific judgement. A promising molecule must still be tested for effectiveness, toxicity, resistance, manufacturing feasibility, and suitability for further development, according to the source article.
Why This Matters for Malaysian SMEs
Most SMEs do not have genome databases or research laboratories. You do, however, face your own “needle-in-a-haystack” problems. A retailer may have thousands of transaction records but no clear view of which products are often bought together. A distributor may have years of WhatsApp messages and invoices but struggle to identify recurring delivery delays. A service company may collect customer feedback but lack the time to classify complaints and find the most common causes.
The same working pattern can help: collect relevant information, ask AI to identify patterns, prioritise the most useful possibilities, and let a person verify the result before acting. For example, you could ask an AI assistant to group customer enquiries by topic, identify unanswered messages, draft follow-up replies in Bahasa Malaysia or English, and produce a short list of issues that require your attention. The AI can accelerate the search; your team remains responsible for confirming facts and making the decision.
Malaysia’s multilingual business environment makes this especially practical. Your staff may communicate with customers in English, Bahasa Malaysia, Mandarin, Tamil, or informal mixed language. An AI assistant can help turn these conversations into consistent categories and summaries. It can also help a team member understand an unfamiliar technical term or create a first draft of a standard operating procedure. The goal is not to make every employee a programmer. It is to lower the barrier between your business knowledge and the digital tools that can apply it.
| Research lesson | SME application | Human check required |
|---|---|---|
| Search a large dataset for useful patterns | Review sales, enquiries, stock, or service records | Confirm the data is complete and correctly interpreted |
| Use AI to brainstorm hypotheses | Suggest reasons for low conversion or repeat complaints | Test the explanation against actual customer and staff evidence |
| Automate code and data preparation | Clean spreadsheets, combine reports, or classify records | Check formulas, duplicates, missing fields, and permissions |
| Prioritise candidates for testing | Rank leads, overdue tasks, stock risks, or process bottlenecks | Review high-impact decisions before taking action |
Start With One Repetitive Business Question
Do not begin with an ambitious “AI transformation” project. Choose one question that your team asks repeatedly and that can be answered using information you already hold. Examples include: Which enquiries have not received a reply? Which invoices are frequently disputed? Which service issues lead to repeat visits? Which products are often out of stock? Which marketing messages produce qualified enquiries?
Prepare a small, clean sample first. Remove unnecessary personal information, separate confidential material, and define what a successful answer looks like. Then ask the AI to explain its method, show assumptions, and identify missing information. If the output affects hiring, credit, customer eligibility, health, safety, or legal compliance, require a human review and keep a record of the final decision.
Use AI to narrow the search and improve preparation—not to skip verification.
A useful workflow has five stages:
- Define: State the business question and the decision it supports.
- Prepare: Gather relevant records and remove unnecessary sensitive data.
- Explore: Ask AI to summarise patterns, alternatives, and possible explanations.
- Verify: Compare the output with source documents, staff knowledge, and real results.
- Improve: Record what worked and update the process when conditions change.
The Bigger Picture
The antimicrobial research story points to a broader change in how specialised work gets done. Previously, a person often needed years of training in several fields before they could move from an idea to a working analysis. Tools such as ChatGPT and Codex can help a small team cross those boundaries more quickly. They can explain unfamiliar concepts, create a first version of a script, restructure information, and make collaboration easier.
That does not eliminate expertise. It increases the value of people who understand your customers, operations, industry risks, and Malaysian business context. An AI system may find a pattern in your sales records, but you know whether a festival period, supplier shortage, public holiday, or local customer habit explains it. The strongest results come from combining machine speed with human context.
For your SME, the competitive advantage may not come from using the most fashionable AI tool. It may come from building a disciplined habit: keep reliable records, ask focused questions, verify important outputs, and turn successful experiments into repeatable workflows. Just as the researchers combine AI predictions with laboratory evidence, you can combine AI analysis with customer conversations, operational checks, and measured business results.
The practical takeaway is simple: start where information is abundant and attention is limited. Let AI help your team find the signal, but make sure your people decide what the signal means and what to do next.
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 →