When Your Business Needs Better Evidence
You may not be running a biotech company, but you face a similar business problem every week: you need to make decisions based on evidence, yet your available information is often incomplete, delayed or too far removed from real customer behaviour.
A product may pass an internal test but perform differently in a customer’s hands. A marketing message may look convincing in a meeting but fail online. A new process may appear efficient on paper while creating problems for your staff. The gap between a controlled test and real-world performance can quietly affect your results.
A recent example from biotechnology shows how one company is trying to close that gap. Outer Biosciences, co-founded by Michael Polansky, is training artificial intelligence models using living human skin kept outside the body. The company says its system can keep donated skin viable for up to a month, compared with an industry norm measured in days, allowing researchers to observe slower processes such as pigmentation changes, collagen remodelling and barrier repair. Source: TechCrunch
TL;DR
Outer Biosciences is using living human skin, ethical tissue sourcing and AI to study how skin changes over time. For your SME, the broader lesson is simple: build systems that capture real behaviour over a useful period, then use structured data to improve decisions.
You do not need advanced laboratory tools to apply this idea. You need better observation, clean records and tests that reflect how your customers and employees actually behave.
What This Means
Traditional biological testing often relies on animal models, simplified cell cultures or lab-grown organoids. These methods can provide useful information, but they may not fully represent how a living human organ behaves. Outer Biosciences is taking a different approach by receiving human skin that would otherwise be discarded after surgery, with the company describing its sources as vetted tissue organisations operating under institutional review board oversight and documented donor consent. Source: TechCrunch
The tissue is delivered within hours of surgery, de-identified by the supplying organisations and placed in a support system that supplies nutrients and removes metabolic waste. According to the company, this keeps the tissue viable for up to one month while preserving important biological characteristics. Source: TechCrunch
That longer observation period matters because some changes cannot be understood immediately. Acute toxicity may be visible quickly, but processes such as skin recovery, pigmentation and barrier repair take longer. The company says its researchers can expose living tissue to UVB damage and observe stress, inflammation and recovery-related responses over the following weeks. Source: TechCrunch
The AI model is not replacing scientists. It helps organise and interpret large amounts of biological information gathered over time. The practical principle is broader than biotechnology: better decisions come from observing the right thing for long enough, then turning those observations into usable information.
“Do not confuse a quick test with a complete test. A result becomes more useful when it reflects the time, conditions and behaviour your customer will actually experience.”
How This Applies to Malaysian SMEs
1. Test your customer journey over time, not just at launch. Suppose you operate a tuition centre, beauty business, online store or service company. You may check whether a customer can submit an enquiry form, make a booking or complete a purchase. That is only the first moment. The more important question is what happens over the following days.
Does the enquiry receive a reply within your promised timeframe? Does the customer receive the correct reminder? Does a new buyer understand how to use the product? Does a client return for a second appointment? By recording these events for several weeks, you can see problems that a one-day test will miss. A simple customer relationship management system can record enquiry date, response time, appointment status, follow-up activity and repeat purchase behaviour.
2. Use real operating conditions when testing internal processes. A standard operating procedure may work when the owner is supervising every step. It may fail when your team is busy, when a staff member is absent or when several orders arrive at once. Test your process during normal working conditions, including peak periods.
For example, a Malaysian food manufacturer can record the time from order confirmation to production, packing and delivery handover. A trading company can track how often purchase orders need correction. A repair workshop can monitor diagnosis time, parts waiting time and customer collection delays. These records show where work slows down instead of relying on assumptions.
3. Build a longer feedback loop for products and campaigns. A social media campaign may produce clicks in its first few days, but clicks do not tell you whether the audience becomes qualified leads or repeat customers. Track the full journey: source of enquiry, response time, quotation status, conversion, fulfilment and repeat activity.
If you sell skincare, supplements or household products, customer feedback may also change after several weeks of use. Do not make claims beyond what you can support. Instead, create a structured feedback process that asks customers about usage, problems, satisfaction and whether they would purchase again. Keep personal data protected and obtain clear permission before using testimonials or customer information.
A Simple Observation Framework
| Business area | What to observe | Useful review period |
|---|---|---|
| Customer enquiries | Response time, follow-up and conversion | 7 to 30 days |
| Online campaigns | Lead quality, completed purchases and repeat actions | 14 to 60 days |
| Operations | Delays, rework, handover errors and completion time | 7 to 30 days |
| Customer retention | Repeat bookings, complaints and referrals | 30 to 90 days |
These review periods are practical starting points, not universal rules. Choose a period that matches your sales cycle. A restaurant may learn quickly from daily records, while a renovation contractor may need several months to understand referrals and project follow-up.
Practical Takeaways
- Define the real outcome first. Decide whether you are measuring faster response, fewer errors, repeat orders or improved customer satisfaction.
- Capture events consistently. Use the same fields and definitions so that different staff members record information in the same way.
- Observe the whole journey. Do not stop at clicks, enquiries or completed forms. Track what happens afterward.
- Review trends, not isolated incidents. One complaint deserves attention, but repeated patterns deserve process changes.
- Keep data clean. Remove duplicate records, standardise customer names and record dates in one format.
- Protect personal information. Collect only what you need, limit access and explain how customer information is used.
- Automate reminders and reporting. Let your system flag overdue follow-ups, unresolved complaints and stalled quotations.
- Run small tests. Change one part of a process at a time so you can identify what caused the result.
- Ask staff for context. Numbers can show that delays exist, while employees can explain why.
What You Can Do This Week
Choose one process that regularly causes frustration. It could be responding to WhatsApp enquiries, preparing quotations, scheduling technicians or confirming deliveries. Write down each step from the first customer request to the final outcome.
Next, select three to five measures. For a quotation process, these could be response time, number of revisions, approval rate, days before follow-up and reasons for lost opportunities. Record them for at least one working week, then review the pattern with your team.
Finally, introduce one small improvement. This might be a standard enquiry form, an automatic reminder, a shared job status board or a required checklist before delivery. Continue observing the same measures. If the result improves, document the new process so it does not depend on one experienced employee remembering every detail.
The Bigger Picture
The important trend is not simply that AI is being used in a biology laboratory. It is the combination of real-world inputs, longer observation and structured analysis. Businesses in every sector are moving away from decisions based only on intuition or short-term snapshots.
For Malaysian SMEs, this does not mean you need to build a sophisticated AI model immediately. It means you should make your business easier to observe. If customer information is spread across personal phones, paper notes and separate spreadsheets, you cannot easily see the full picture. If every staff member describes a “completed job” differently, your reports will remain unreliable.
Start with consistent data and clear workflows. Once those foundations are in place, automation tools can help you identify overdue tasks, compare performance and spot recurring issues. AI may eventually assist with predictions and recommendations, but its usefulness will depend on the quality and relevance of the information you provide.
The living-tissue example offers a useful reminder: a model becomes more valuable when it is trained on conditions that resemble reality. Your business systems should follow the same principle. Measure what customers actually experience, watch what happens after the first interaction and use the findings to make one practical improvement at a time.
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 →