Why EXL’s AI Ambition Needs Clean Data and Solid Operations

I love a confident CEO, but confident slides don’t clean up messy data.

EXL’s recent investor day reads like a checklist many of us want to see: AI, data management and operations positioned as pillars for growth. That’s a sensible position. Models matter, but for enterprise customers the difference between a successful deployment and an expensive experiment usually comes down to data quality, decision hygiene and operational readiness.

I’m cautiously optimistic about EXL’s AI push. They’ve got things most AI vendors lack: domain know‑how in verticals like insurance and financial services, real data engineering chops, and experience running operations at scale. Those capabilities are precisely the parts of the value chain that make automation sustainable. But cautious optimism is the key phrase. Double‑digit growth promises are only as real as the plumbing behind them.

I’ve seen this movie before. A midsize insurer I worked with bought a commercial AI claim‑routing tool and expected overnight magic. What they actually got was chaos: duplicate customer records across systems, a hodgepodge of undocumented edge‑case rules accumulated over years, and front‑line agents who didn’t trust the routing recommendations. The vendor’s model was fine on paper, but because inputs were noisy and decisions were opaque, the business couldn’t lean on the system.

The invisible work — the months of deduplication, rule rationalisation, process simplification and retraining — is precisely what EXL’s pitch implicitly acknowledges. Data management and operations experience matters because that’s the unglamorous, tedious work that turns automation into measurable value. Without it, AI becomes an expensive band‑aid that amplifies problems rather than solving them.

That said, don’t throw shade on the ambition. Winning enterprise customers actually requires solving messy problems at scale. Firms that combine domain expertise with operational delivery are rare and valuable. My worry isn’t that EXL overestimates AI’s potential; my worry is that enterprise adoption timelines are long, integration complexity compounds quickly, and competition on price and delivery will be fierce. The label “strategic trusted partner” only holds up if trust is proven in the small print: solid SLAs, pragmatic onboarding, and committed post‑launch troubleshooting.

If you’re a business watching this and thinking about following suit, here are four practical steps to take first:

1) Fix the basics. Master your data: eliminate duplicates, reconcile sources and codify decision rules. Garbage in, garbage out still holds.

2) Start tiny. Run a narrow pilot that addresses a single, measurable pain point for one team. Don’t boil the ocean on day one.

3) Measure the right things. Track time saved, error rates, customer satisfaction and downstream impacts — not just model accuracy or vanity metrics.

4) Design for human‑in‑the‑loop. Use AI for repetitive, low‑risk tasks and keep humans where judgment matters. That preserves trust and creates clearer escalation paths.

I’m rooting for firms like EXL — businesses that understand AI without operations is like a sports car without fuel. Their pitch reads like a company that values the grunt work: the data engineering, the process design, the operational playbooks. For executives chasing double‑digit numbers, the smart bet is less about buying the flashiest model and more about fixing the driveway so the car can actually drive.

If you want help figuring out which potholes to patch first, I’ve got a list and a couple of war stories — and yes, I’ll bring the coffee.

Source: [ExlService Investor Day Spotlights AI Push, Double‑Digit Growth Goals](https://www.americanbankingnews.com/2026/05/18/exlservice-investor-day-spotlights-ai-push-double-digit-growth-goals.html)