When “Coming Soon” Is Not a Business Strategy

You may have experienced this in your own business: a software vendor promises that an important feature is “coming soon”, so you plan around it. Months later, your staff are still using spreadsheets, manual reminders, and workarounds. The promised improvement may eventually arrive, but your daily operations cannot depend on an uncertain date.

That is the practical lesson behind the latest discussion about Tesla’s Full Self-Driving system and its promised ability to avoid potholes. According to Electrek, Elon Musk again described pothole avoidance as “coming soon”, despite a similar promise being reported in 2019. The issue is not only whether the feature eventually works. It is whether people can safely and sensibly rely on a promise before the result has been proven.

For you as a Malaysian SME owner, this matters because business automation also sits on a spectrum between assistance and autonomy. A system can recommend an action, carry out a routine task, or make a decision with little human involvement. The more responsibility you give software, the more carefully you need to test its reliability.

TL;DR

Do not automate critical business decisions based on a vendor’s future roadmap. Start with measurable, repeatable tasks, test the system using your own data, and keep a clear human approval step until performance is dependable.

The goal is not to remove people from every process. The goal is to make routine work more consistent while giving your team better information and control.

What This Means

A pothole is a useful example because it looks simple to a human driver but requires several steps from an automated system. The vehicle must identify the hazard, decide whether it is safe to move around it, check nearby traffic, select an alternative path, and respond correctly if the road condition changes. A system may perform well on ordinary roads while still making mistakes in unusual or difficult situations.

The same pattern appears in business software. Consider an automated invoice process. It may read the supplier name correctly but misunderstand a tax amount. It may match a purchase order but fail to notice that the quantity is wrong. It may send a reminder to a customer whose payment has already been made. Automation is not simply about recognising information. It must also make a suitable decision in context.

The Electrek article describes progress in Tesla’s driver-assistance software, while also highlighting the gap between broad capability and specific promised features. It reports that Tesla’s systems can navigate many roads and perform certain parking tasks, but still struggle with common road hazards such as potholes. The article also notes that Tesla has made earlier predictions about self-driving capabilities that did not arrive within the expected timeframe. Read the full reporting at Electrek.

Useful automation is not defined by how impressive the promise sounds. It is defined by how reliably the system handles your real work today.

How This Applies to Malaysian SMEs

Imagine you run a wholesale distribution business in Selangor. Your sales team receives orders through WhatsApp, email, and phone calls. An automation tool may claim it can extract product details and create orders automatically. That could save time, but Malaysian product names, abbreviations, mixed languages, and informal messages can create errors. For example, “2 carton mineral” may not specify the exact pack size. The safer approach is to let the system prepare a draft order while a staff member confirms the product, quantity, delivery address, and requested date.

Now consider a service company in Penang that sends technicians to customer sites. An automated scheduling system can assign jobs based on location, availability, and skill. However, it may not understand that a technician needs a particular part in the van, that a customer only allows access after office hours, or that heavy rain could affect travel time. Use automation to organise the schedule and highlight conflicts, but give your operations person authority to adjust it. This keeps the speed of software without pretending that every local detail can be predicted.

A small restaurant, retailer, or online seller faces a similar issue with inventory. A system can identify fast-moving items and suggest replenishment. But a sudden school event, festive season, promotion, or supplier delay may change demand quickly. You should review unusual recommendations instead of allowing the system to reorder everything without checks. The automation can monitor stock levels every day, flag products below a threshold, and prepare a purchase list. Your team still decides what makes operational sense.

For professional firms such as accountants, agencies, and consultants, automated document handling can classify files, extract dates, and remind clients about missing information. Yet confidential documents require careful access controls. A tool that saves administrative time but sends information to the wrong recipient creates a larger problem. Before adopting it, confirm where data is stored, who can access it, how activity is logged, and how your team can correct mistakes.

Assistance Versus Autonomy

It helps to separate three levels of automation. At the first level, the system observes and reports: it tells you that a payment is overdue or that stock is low. At the second level, it recommends an action: it drafts a reminder or suggests a purchase quantity. At the third level, it acts automatically: it sends the reminder, places the order, or changes a customer record.

Automation level	Example for an SME	Suitable control
Monitor	Flag overdue invoices or low stock	Staff reviews the alert
Recommend	Draft a customer message or delivery schedule	Staff approves before sending
Execute	Send reminders or update a workflow automatically	Set rules, limits, logs, and exception handling

The right level depends on the consequences of an error. An incorrect internal reminder may be easy to fix. An incorrect payroll record, customer quotation, stock order, or compliance document deserves stronger review.

Practical Takeaways

• List the process before choosing the tool. Write down the current steps, who performs them, where delays occur, and what usually goes wrong.
• Start with repeatable work. Good early candidates include appointment reminders, lead assignment, document filing, simple approval routing, and daily status reports.
• Define success with your own measures. Track completion time, correction frequency, missed follow-ups, and staff adoption. Do not rely only on a product demonstration.
• Ask what happens when the system is unsure. A reliable workflow should send an exception to a person rather than silently guessing.
• Keep an approval step for sensitive actions. This is especially important for payroll, customer data, tax-related records, supplier orders, and external communications.
• Request evidence, not only promises. Ask for a working demonstration using a realistic sample of your documents or messages.
• Record every automated action. Activity logs help you identify errors, explain what happened, and improve the workflow.
• Review performance regularly. A process that works during normal weeks may need different rules during promotions, festive periods, or staff shortages.

A Simple Test Before You Automate

Choose one process that happens frequently and has a clear beginning and end. For example, handling a new enquiry. Measure how many enquiries arrive each week, how long it takes to respond, how many are missed, and how often staff enter incomplete information. Then introduce automation in stages: capture the enquiry, assign an owner, prepare a response, and only later consider automatic sending.

Run the workflow with real examples, including messy messages and incomplete details. Ask your staff to note every correction. If the system performs well, expand its responsibility gradually. If it fails, you have limited the damage and learned exactly what needs improvement.

The Bigger Picture

The long-term direction is clear: software will handle more routine observation, prediction, and action. However, greater capability does not mean every task should be fully automatic. Physical roads contain unexpected hazards, and businesses contain equally unpredictable human situations: unclear requests, changing priorities, exceptions, and relationships that require judgement.

For Malaysian SMEs, the strongest approach is practical rather than flashy. Build dependable workflows around the way your business actually operates. Connect the tools your team already uses, keep information organised, and make responsibilities visible. Automation should help your people notice issues earlier and complete routine work consistently.

When a vendor says a major capability is coming soon, treat that as a possibility, not a plan. Plan around features that are available, tested, and supported now. Then you can adopt future improvements when they are proven, without putting your customers or operations at risk.

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 →