Will AI Replace BPO?

AI Can Replace 85% of BPO’d Work - Here’s Why That’s Good for Customers, Employees, and Compliance

The big shift

For two decades, BPO has been the duct tape for repetitive, rules-based work. In 2024, IT & telecom, BFSI, healthcare, retail, and manufacturing were the heaviest BPO users - outsourcing everything from tech support and helpdesk to mortgage processing, claims admin, finance & accounting, and omnichannel customer care.

Even with macro headwinds, global BPO spend remains massive - approximately $302.6B in 2024 and tracking to $328B+ in 2025 (methodologies vary; some models place 2025 higher).

But the ground is moving. Generative AI entered the contact center in a “big way” in 2023–2024 and is now steadily absorbing the exact classes of work that BPO scaled so well: high-volume, text- and speech-driven, governed by policies and knowledge bases.

Why 85% is a reasonable, if provocative, hypothesis

What BPO does best: clear rules, high volume, narrow scope, measurable SLAs, and limited need for deep context. That’s also exactly where today’s AI agents and automation excel - especially when paired with retrieval over enterprise data, action connectors, and guardrails.

• Contact centers are the leading indicator. Mainstream reporting shows AI already reshaping roles and cutting some jobs; industry leaders publicly forecast steep reductions in traditional call-center work as agents shift to exceptions and escalations.

  
  

• Enterprise platforms are shipping “digital labor.” Salesforce Agentforce positions autonomous agents to handle frontline service/sales tasks and orchestrate cross-system actions - with governed handoffs to humans. That is a direct alternative path to sending work offshore.

  
  

• Market signals corroborate the pivot. Everest-tracked BPO growth slowed in 2023 while automation accelerated; Gartner/Reuters note the agentic rush (with expected shake-outs), but still expect agentic capabilities to pervade enterprise software by 2028.

  
  

Taken together: if you decompose BPO portfolios into repeatable intents and workflows, then equip AI agents with retrieval, tool use, and policy guardrails, the automatable portion is large. In many shops, 85% of ticket or case volume is routine (status checks, data updates, simple eligibility decisions, document intake) that can be handled end-to-end by AI with human review on the tails. (McKinsey repeatedly documents the scale of productivity and containment gains when this is done well.)

A Familiar Example: On-shore AI-first beats offshore BPO, especially in recruiting

Staffing and recruiting is a perfect microcosm. Employers adopting AI assistants (screening, scheduling, FAQ, assessments) are compressing cycle times dramatically. Nestlé and Pfizer have public case studies using Paradox’s Olivia to automate candidate screening and interview coordination.

Industry surveys and reports show a growing share of HR teams using AI to source, pre-screen, match, and schedule, cutting time-to-hire and reducing manual touches that once justified BPO headcount.

Why this matters: when you keep the workflow on-shore with AI, you preserve data residency, improve auditability, reduce handoffs, and protect brand experience while often lowering cost more than offshore BPO after the first few months of agent training and integration.

The product wave: niche AI for every operational function

• Customer support now has credible AI agents that resolve routine issues across chat, email, and voice, with clean human handoff. Reviews and analyst coverage suggest broadening adoption and maturing quality.

  
  

• Insurance underwriting is being re-tooled with AI that ingests submissions, scores risk, detects misrep/fraud, and recommends terms, yielding materially faster reviews and tighter loss selection; EY’s 2024 study confirms widespread GenAI experimentation by carriers across underwriting and policy servicing.

  
  

Expect similar dynamics in F&A, claims, KYC/AML checks, invoice processing, order management, and field-service dispatch as intelligent automation platforms and agent frameworks standardize discovery → build → run.

The nuance: not “AI replaces BPO overnight”

Gartner warns that >40% of agentic AI projects could be scrapped by 2027 due to immaturity and unclear ROI even as agentic capabilities spread widely by 2028.

Translation: the direction is certain, the cadence is uneven, and governance/data quality will separate winners from re-branders.

Meanwhile, reputable market trackers still show hundreds of billions in BPO outlay through 2025, so incumbents that reinvent around automation and high-skill exception work will remain relevant.

What leaders should do now

1. Inventory BPO scopes by “automation-readiness.” Classify intents by rule clarity, data accessibility, exception rate, and regulatory constraints.

   
   

2. Stand up an agentic runway (retrieval, tools, guardrails) in your core platforms (e.g., Salesforce) to move high-volume intents on-shore into AI agents with human-in-the-loop.

   
   

3. Target a 3 to 6 month payback: start with 2 to 3 “boringly valuable” intents (password resets, order status, eligibility lookups, document ingestion), then scale. McKinsey’s data shows value concentrates where you pair AI with process redesign and robust KM.

   
   

4. Keep a human exception lane for identity theft, complex grievances, and regulated adjudications - areas that still need people.

   
   

APG’s Position

We believe 85%+ of today’s BPO’d tasks can transition to AI-led “digital labor” with the right method, platform, and controls. Here’s how we make it real:

• Method: APG’s AI Microbursts™ - a rapid, disciplined approach to discover, design, and deploy small, high-ROI automations that compound into meaningful OPEX reduction and CX lift.

  
  

• Platforms: We build on Salesforce Agentforce & Einstein for governed, enterprise-grade AI agents; we also leverage OpenAI and Anthropic (for reasoning and language), and Bubble (for fast, compliant front-end workflows).

  
  

• Principles: Lean and fast without sacrificing quality, employee experience, customer experience, or regulatory and legal compliance. We design for auditability, encryption, RBAC, PII minimization, and human-in-the-loop for sensitive decisions (e.g., underwriting, adverse actions).

  
  

Bottom line: If your BPO scope is largely rules-based and data-retrieval heavy, AI can do the work faster, cheaper, and more controllably on-shore. The sooner you start converting tickets and cases into governed AI workflows, the sooner you’ll exit the “volume treadmill” that BPO optimized and AI can now eliminate.

Ready to discuss AI and automation replacing some or all of your BPO expenditure? Contact us today.