AI Tool Choices Are Moving Faster Than Your Business
If you run a Malaysian SME, you may already feel pressure to “do something with AI”. Your competitors are testing chatbots, sales teams are drafting messages with AI, and staff are asking which assistant they should use for documents, coding, customer service, or research.
The difficult part is not finding an AI tool. There are plenty. The difficult part is choosing one that your team can use consistently, safely, and productively—without changing direction every time a new model is announced.
Recent business-use data from Ramp shows why this decision is becoming less straightforward. Among more than 70,000 United States businesses using Ramp’s payment and expense-management products, Anthropic held about 44% of paying business users in July, compared with nearly 40% for OpenAI. The same data indicated that OpenAI was growing faster during the third quarter at that point. Source: TechCrunch
TL;DR
Business AI adoption is growing, but companies are still switching between providers as models and policies change. For your SME, choose tools based on repeatable workflows, data controls, staff adoption, and reliable results—not brand popularity alone.
Start with one or two practical processes, measure the outcome, and keep your business information portable so you are not trapped by one provider.
What This Means
The simplest way to understand the news is this: businesses are not treating AI assistants as permanent choices yet. They are testing different tools and moving when another provider offers better results, stronger controls, or a better fit for a particular job.
OpenAI was once widely seen as the leading choice for both consumers and businesses. However, Ramp’s figures showed Anthropic moving ahead among its paying business customers in May, with 41% compared with OpenAI’s 39%. By July, Anthropic had nearly 44% while OpenAI had nearly 40%. Source: TechCrunch
These figures do not represent every company or the entire global market. Ramp’s customers are mainly American businesses, and its customer base has a technology-industry skew. The data also excludes businesses using other corporate payment and expense systems. Source: TechCrunch
Still, the movement is useful for one reason: it shows that business AI selection is still fluid. A company may prefer one tool for writing, another for coding, and a third for handling internal documents. Your decision should therefore be based on the work you need completed, rather than trying to identify one “best” AI provider for everything.
The right question is not “Which AI brand is winning?” It is “Which workflow becomes more reliable when my team uses this tool?”
How This Applies to Malaysian SMEs
For a Malaysian retailer, distributor, or service business, AI can support everyday work such as preparing product descriptions, replying to common customer questions, summarising supplier emails, and converting meeting notes into task lists. However, the value comes from fitting the tool into your existing process. If staff copy information manually between WhatsApp, spreadsheets, email, and an AI assistant, mistakes can still happen. You need a clear workflow showing what goes in, what the AI may do, and who checks the final output.
Consider a small trading company serving customers in Bahasa Malaysia and English. Your team may use AI to draft quotations, translate product explanations, or prepare follow-up messages. The tool must understand local phrasing, product terms, and your preferred tone. A polished answer that gives the wrong delivery condition or misunderstands a Malaysian business expression is not useful. Test the tool with real but non-sensitive examples before allowing it into customer-facing work.
For an accounting, property, recruitment, or professional-services firm, information handling becomes even more important. Your staff may work with identity documents, employee records, customer details, contracts, or financial information. Do not paste confidential material into a personal AI account simply because it is convenient. Create rules for what information may be entered, remove unnecessary personal details, and require human review for advice, calculations, compliance statements, and commitments made to customers.
Manufacturers and wholesalers can also use AI for internal knowledge access. You could organise standard operating procedures, maintenance instructions, product specifications, and frequently asked questions so staff can find answers faster. The important point is to connect the assistant to approved documents, not allow it to invent company policy. When information changes, assign one person to update the source documents and record the revision date.
AI provider changes may also affect your operations. A model can become better at one task, slower at another, or subject to different data-retention rules. Anthropic’s warning that certain user data might need to be retained for 30 days illustrates why policy details matter alongside model quality. Source: TechCrunch Before adopting any tool, check its business terms, administrator controls, access permissions, retention practices, and export options.
Useful Numbers from the Report
| Indicator | Reported figure | Why it matters |
|---|---|---|
| Ramp business customers covered | More than 70,000 | Shows the dataset is substantial, although it is not the whole market. |
| Anthropic share in May | 41% | Indicates Anthropic moved ahead among paying Ramp business users. |
| OpenAI share in May | 39% | Shows the gap was narrow rather than a permanent separation. |
| Anthropic share in July | Nearly 44% | Suggests Anthropic retained a lead in that dataset. |
| OpenAI share in July | Nearly 40% | Shows OpenAI remained close and was reportedly growing faster in the following quarter. |
| Ramp companies paying for AI in July | Nearly 56% | Indicates business adoption was expanding across the dataset. |
All figures in this table come from the TechCrunch report based on Ramp data. Source: TechCrunch
Practical Takeaways for Your Business
- Choose a workflow first. Start with one problem, such as replying to enquiries, preparing summaries, or searching internal procedures.
- Run a short test with real scenarios. Use several common examples, including difficult cases, instead of judging the tool from one impressive demonstration.
- Set an owner. One person should manage prompts, approved documents, access rights, and feedback from staff.
- Keep a human approval step. AI-generated quotations, legal wording, financial calculations, hiring messages, and customer promises should be checked before sending.
- Protect confidential information. Define which data staff may enter and whether personal information must be removed first.
- Measure useful outcomes. Track response time, rework, unanswered enquiries, document-search time, or completed tasks—not just how often staff open the tool.
- Avoid one-tool dependency. Keep important procedures and source data in systems you control, with clear export and backup arrangements.
- Review the arrangement regularly. Recheck quality, usage, access, retention policies, and staff feedback every few months.
A Simple Starting Plan for Malaysian SMEs
- List five repetitive tasks. Ask your team where they spend time copying, summarising, rewriting, or searching.
- Select one low-risk task. Avoid starting with sensitive customer or employee data.
- Create a standard instruction. State the context, expected format, language, tone, and information the AI must not invent.
- Compare results against your current process. Check accuracy, time saved, and the amount of correction required.
- Document the approved process. Save the prompt, example output, reviewer responsibility, and escalation rule.
- Expand only after adoption. If staff do not use the first workflow, another tool will not solve the underlying process problem.
The Bigger Picture
The long-term lesson is that AI providers will continue competing for business users, and their relative strengths may change quickly. The report also indicated that the share of Ramp customers paying for AI rose from more than 50% in March to nearly 56% in July. Source: TechCrunch That suggests the broader business market is growing even while companies move between providers.
For you, this means AI planning should look more like process management than software shopping. Build good instructions, clean up your internal documents, establish access rules, and teach staff how to check outputs. Those capabilities remain useful even if you change providers later.
You do not need to predict which AI company will lead next year. You need to know where your business loses time, where errors are common, and where a controlled assistant can help your team work more consistently. Start small, keep the results measurable, and make your business process strong enough to survive a change in tools.
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 →