Will AI Replace My Employees? Here's the Honest Answer

93.9 percent. That is the share of Indonesia's administrative and clerical workers whose jobs overlap with generative AI, while 67.5 percent of all clerical workers fall into the highest exposure category (Source: ILO, 2026). If your admin team spends every day copying Shopee and Tokopedia orders into a spreadsheet or chasing QRIS transfer receipts, it's natural for a number like this to give you pause.

Here is the honest answer for business owners asking whether AI will replace jobs: it takes over work one task at a time, and rarely a whole role, at least in the short term. For most employees, the work itself will change. The real risk sits in entry-level hiring for roles like admin and customer service: new seats that never get filled, not long-serving staff being let go.

What does the data say about the risk?

A high exposure figure doesn't mean the seat will sit empty. The same ILO study found that across the ASEAN countries it examined, 21 to 28 percent of jobs have some exposure to generative AI. The highest-exposure group, which the ILO links to an elevated risk of job displacement, accounts for only about 3 to 4 percent of jobs in Indonesia, the Philippines and Thailand, and under 2 percent in Vietnam (Source: ILO, 2026). Most workers face partial automation of their tasks.

The ILO's global research paints a similar picture. After scoring nearly 30,000 tasks, it concluded that most jobs will be reshaped rather than eliminated, because they still need human input. The average automation score dipped slightly, from 0.30 in 2023 to 0.29 in 2025 (Source: ILO, 2025). The exception is media and web work, where scores rose as AI got better at handling voice, images and video.

One point worth flagging if you usually hire fresh graduates. In Indonesia, 26.1 percent of young workers aged 15 to 24 are in jobs exposed to generative AI, compared with 21.1 percent of adult workers (Source: ILO, 2026).

Where is AI replacing jobs first?

The most detailed data so far comes from the United States, and the pattern is telling. Stanford's Digital Economy Lab analysed US payroll data through June 2026 and found no evidence of widespread job displacement. Among 22- to 25-year-olds in the most AI-exposed occupations, employment is now about 19 percent below where it would be had it kept pace with same-age workers in less-exposed jobs, a shortfall that has widened from 15 percent in July 2025 (Source: Stanford Digital Economy Lab, 2026).

The cause is what matters to you. That shortfall comes mainly from companies hiring fewer people, not from more employees leaving. Experienced workers show nothing comparable. The effect is also concentrated in jobs where AI substitutes for tasks; in jobs where AI assists people, employment has held flat or even grown (Source: Stanford Digital Economy Lab, 2026).

Research from Anthropic points the same way. Customer service and data entry are among the occupations whose tasks AI is most often doing in practice, with 67 percent coverage for data entry keyers. For US 22- to 25-year-olds in exposed occupations, the job-finding rate fell about 14 percent after ChatGPT arrived, while unemployment among the most exposed workers has not risen significantly (Source: Anthropic, 2026).

What about companies that really did cut staff?

Honesty means looking at the counterexamples too. Starting in early 2025, Salesforce cut its customer support team from about 9,000 to about 5,000 people: roughly 4,000 roles gone. CEO Marc Benioff said AI agents now handle about half of customer conversations and support costs fell 17 percent. A company spokesperson said only hundreds of affected staff were redeployed into professional services, sales and customer success instead of being laid off (Source: Fortune, 2025).

Klarna offers a lesson from the other direction. In February 2024, Klarna reported its AI assistant had handled 2.3 million chats in its first month, about two-thirds of service inquiries, which it equated to the work of 700 agents. The company had laid off staff and paused hiring. In May 2025, CEO Sebastian Siemiatkowski admitted that cost had weighed too heavily in that decision, and Klarna started recruiting human agents again under a flexible remote-work model (Source: CX Dive, 2025).

In a WhatsApp-first market like Indonesia, Klarna's lesson cuts even deeper. Customers expect a human to reply. In our view, sending AI replies straight to customers with nobody checking them is the fastest way to damage relationships that took years to build.

What this means for your business

Start by seeing an admin role as a bundle of tasks you can pull apart. Answering repetitive chats, recapping orders and re-typing invoices or bank statements are highly exposed. Calming an angry customer, deciding on a refund, judging an exception and negotiating with suppliers are not. Part of that work will go to AI, and the role shifts with it.

Across the systems we build, the human work moves from typing to checking, correcting and handling exceptions. In Finance AI Automation, documents that used to be re-typed every month are now read by the system, yet nothing is booked without human confirmation, and any amount missing from the source document is blocked, never rounded. Built by Agentiq for an anonymised client; all screens shown use fictional sample data.

In Olaif Engage, AI prepares draft replies, and a reviewer approves, edits or escalates them before anything is sent. Olaif Engage is a real project built by Agentiq for Olaif; its case study uses fictional sample data, and every figure in it is illustrative.

If cutting staff has crossed your mind, understand the legal framework first. Indonesia's Job Creation Law, Law No. 6 of 2023, requires that termination be avoided wherever possible, and if it still happens, the employer must pay severance and/or long-service pay plus compensation for entitlements (Source: Law No. 6 of 2023). For a specific case, talk to your HR team or legal counsel.

Short of layoffs, the more realistic options are to move employees into approval work, exception handling and customer relationships, or to leave a seat unfilled when someone resigns. A third option is to absorb business growth without adding new admin staff.

The biggest constraint, it turns out, is skills. A survey by AWS and Strand Partners of 1,000 business leaders and 1,000 members of the public in Indonesia estimates that 48 percent of future jobs will need AI literacy, yet only 21 percent of businesses feel their workforce is ready, and 57 percent name a lack of skilled personnel as the leading barrier (Source: AWS/Strand Partners, 2025). We believe training the admin who already knows your business is generally cheaper and less risky than recruiting a dedicated AI specialist.

Steps you can start this week

• Map out the daily tasks of each admin and customer service person, then separate repetitive work from work that needs judgment.
• Before you advertise your next admin vacancy, check whether a system could take on part of the work.
• Design every automation with a human approval step, and train your existing people to take on that reviewer role.
• Bring in HR or legal counsel before making any decision that affects an employee's status.

If you'd like to see how the work could be split between AI and your own team, walk us through the process on our contact page. The Agentiq team will reply with concrete recommendations, not a template.