About 40,000 Jamaicans work in business process outsourcing, and for many young people it is their first formal job. AI will cut part of that work. This article goes role by role through what is exposed, on what timeline, and what an individual worker can do about it. For the news behind the sector's recent job losses and Digicel's Ruby assistant, see my separate piece on that; this one is the practical assessment for people inside the industry.
The sector today
By the industry's count, employment stands at roughly 40,000, down from about 60,000 two years earlier. Annual receipts were close to US$1 billion at their peak and fell to US$780 million in the financial year to March 2026. Operators have included itel, Conduent, Hinduja Global Solutions and Sutherland, concentrated in the Kingston Metropolitan Area and Montego Bay.
Weekly pay runs from about J$40,000 for entry-level agents to J$120,000 for specialised technical roles. The sector's pitch to US clients has been English-speaking staff in a nearby time zone at roughly 40% to 60% less than the same agent in Ohio or Texas. AI competes on exactly that price.
What AI already does at production quality
- Routine calls and chats: order status, balances, password resets, appointment booking and basic troubleshooting. Industry estimates put an automated interaction at cents, against several dollars for a human-handled one.
- Document processing: reading, classifying and extracting data from claims, invoices and standard forms.
- Data entry and validation from clean, structured sources.
- Email and chat replies drafted from a company's knowledge base and customer records.
These tools are in use, not in development. What decides the pace in Jamaica is how fast clients move work from human agents to software, and that depends on their own pilots, regulators and contracts.
Risk by job category
The headcounts below are my estimates for how the sector's workforce splits across roles. The timeline is when I expect a noticeable effect on hiring.
| Role | Est. workers | Risk | Timeline |
|---|---|---|---|
| Entry-level voice customer service | 15,000 to 18,000 | High | Under way; 2 to 3 years |
| Data entry and document processing | 5,000 to 7,000 | High | 1 to 3 years |
| Email and chat support | 4,000 to 6,000 | Medium to high | 2 to 4 years |
| Complex service and escalations | 5,000 to 8,000 | Medium | 3 to 5 years |
| Quality assurance and training | 2,000 to 3,000 | Lower | 3 to 5+ years |
| Specialised KPO (medical coding, legal review, financial analysis) | 3,000 to 5,000 | Lowest | 5+ years |
For routine voice work the arithmetic is direct: if software handles 40% of routine calls, a client needs about 40% fewer agents on those calls. Escalation agents keep the cases that need judgement and patience, but their volume drops as the easy cases disappear. In QA, AI monitoring can review every call where a human team typically samples a small percentage, so fewer analysts cover more.
KPO roles are safer because they need certification and domain knowledge, though AI will let fewer people handle the same volume there too.
What is happening in other outsourcing countries
The McKinsey Global Institute estimated that up to 30% of work hours worldwide could be automated by 2030, with customer service and data processing among the most exposed. The World Bank has warned that developing economies built on low-cost outsourcing are especially exposed.
The Philippines, with more than 1.7 million outsourcing workers, is the closest comparison. Its industry body has made AI skills training a priority, and firms there report moving agents from routine queries to complex and back-office work rather than growing headcount. India's large IT and outsourcing firms have said publicly that they are shifting staff from routine tasks to AI-assisted services. I expect Jamaica to follow the Philippines with a lag of one to two years.
Why the sector will shrink but not disappear
- AI creates operating work. Someone has to review bot conversations, maintain knowledge bases, label training data and handle the cases the system cannot. These roles pay better than entry-level calls and can sit in the same buildings.
- Clients move slowly. Enterprises pilot, validate and scale over years, and regulated industries such as healthcare, insurance and banking require human involvement in some steps.
- Jamaica's advantages persist. Fluent English, cultural familiarity with US customers and the same time zone still matter for difficult conversations.
- A cheaper service can reach new clients. Firms that could not afford an all-human outsourced team may buy a mixed AI and human one.
My forecast for the next five years is a further fall of 15% to 25% in sector employment, to roughly 30,000 to 34,000 workers, with higher average pay among those who remain. New roles will offset part of the loss, not all of it. The workers most exposed are entry-level agents in routine calls and data entry who do not retrain. The part of this forecast I am least sure of is timing: a single large client switching to AI could move it forward by a year.
What government and employers should do
JAMPRO and HEART/NSTA Trust should run a transition programme specific to BPO: funded retraining for displaced workers, incentives tied to upskilling, and help for operators moving to AI-assisted services. Employers should train agents on AI tools before cutting roles, not after. And JAMPRO and GSAJ should sell Jamaica to clients on service quality with AI-assisted staff rather than on price alone.
Prompt templates for BPO workers and team leaders
I work as a [specific BPO role] in Jamaica handling [type of work] for [type of client]. Analyse how vulnerable my specific role is to AI automation over the next 1, 3, and 5 years. Then create a personalised upskilling plan that I can follow in my spare time using free resources. The plan should focus on skills that will make me valuable in an AI-augmented BPO environment. Include specific courses, certifications, and timeline.
Act as a BPO industry strategist. I manage a team of [number] customer service agents in Jamaica. Our primary work is [describe work]. Help me develop a transition plan to move my team from purely human service delivery to an AI-augmented model. Include: which tasks to automate first, how to retrain team members, what new roles to create, and how to maintain service quality during the transition.
I am a BPO worker in Jamaica earning J$[amount] per week. I have [X years] of experience in [specific area]. I am concerned about AI affecting my job. Help me create two career development paths: (1) staying in the BPO industry but moving to a more AI-resistant role, and (2) transitioning my BPO skills to a different industry in Jamaica. For each path, give specific steps, timeline, required training, and expected salary range.
What a BPO worker should do this quarter
- Every agent: for one week, tally your contacts by type. If more than half are balance checks, resets, order status or similar, your role is in the high-risk rows above; start step 2 this month.
- Agents in high-risk roles: pick one specialisation and enrol: medical coding (CPC certification), accounting support (ACCA's accounting technician route), IT support (CompTIA A+), or a HEART/NSTA Trust data or software support course. Expect three to twelve months part time.
- Anyone without data skills: complete a free or low-cost data analysis course, such as Google's Data Analytics certificate (financial aid is available on Coursera), and build one small project using your own team's anonymised metrics.
- Agents who want to stay in the industry: ask your team leader to put you on any AI pilot, as a reviewer of bot conversations or knowledge base editor. Those are the roles that grow when routine calls shrink.
- Team leaders and site managers: use the second prompt above to draft a transition plan for your team and present it to your operations manager before the next client contract renewal.
Frequently Asked Questions
How can I tell whether my BPO account is about to be automated?
Watch for three signs: the client asks your team to tag or categorise every contact by reason, a chatbot or IVR is added in front of your queue, and your site stops hiring for that line of business while volumes hold steady. None of these means layoffs are certain, but together they usually mean the client is building the data and the business case for automation.
Is medical coding a safe move for a Jamaican BPO agent?
Safer than routine calls, not immune. Coding for US health insurers requires certification, most often the AAPC's CPC, and knowledge of US rules that keeps humans in the loop for audits. AI already suggests codes, so the coders who last are those who can review and correct AI output quickly and handle complex cases.
Do I lose redundancy pay if I leave before my job is cut?
Yes, in most cases. Under Jamaica's Employment (Termination and Redundancy Payments) Act, redundancy pay goes to employees with at least 104 weeks of continuous service who are dismissed because of redundancy; resigning generally forfeits it. If you plan to leave for training, check first with the Ministry of Labour and Social Security or a union representative.
Which BPO roles are growing because of AI?
Conversation reviewers who check bot transcripts, knowledge base editors who keep answers accurate, workforce analysts who model mixed human and AI staffing, and escalation specialists who take the cases bots hand off. Few job ads use these titles yet, so look for postings that mention chatbot, virtual agent, knowledge management or conversation design.
Can I use ChatGPT or Claude to help with my BPO work?
Only within your employer's policy. Pasting customer names, account numbers or health details into a personal AI account can breach your client's contract and Jamaica's Data Protection Act 2020. Use personal AI tools for practice, study and drafting with made-up data, and ask whether your company has an approved tool for live work.