BBVA deployed specialised generative AI copilots across its Italian and German contact centres to reduce average handling time for routine customer enquiries by over 15%. The assistants, built on BBVA's proprietary orchestration platform and powered by OpenAI's models, consolidate fragmented internal knowledge sources into single-prompt responses delivered in seconds. Rather than replacing agents, the system frees them from information retrieval tasks—previously requiring searches across multiple product manuals and internal documents—allowing them to focus on complex cases requiring personalised analysis. The tool processes roughly 40,000 agent interactions monthly across both markets, with over 260 agents using it daily, whilst maintaining strict data governance by excluding access to customer personal information and requiring human review before any response reaches customers.
The implementation reveals a critical operational insight for CX teams: the most impactful AI deployments emerge from bottom-up identification of friction points rather than top-down technology mandates. BBVA's contact centre teams identified the problem themselves, which likely explains both the rapid adoption rate and the measurable 15% efficiency gain—metrics many enterprise CX initiatives struggle to demonstrate. The governance framework BBVA applied, including mandatory human oversight, prompt quality reviews, and impact assessments, establishes a template for risk mitigation that should inform how teams evaluate vendor solutions like Salesforce Service Cloud or Zendesk's emerging AI features. Yet the question remains whether this model scales beyond high-volume, FAQ-heavy interactions: does a 15% improvement in handling time translate meaningfully for teams managing complex B2B support or omnichannel scenarios where context-switching and knowledge fragmentation operate differently?
The training component—dedicated sessions teaching agents to formulate natural language prompts and provide context—underscores an often-overlooked implementation cost. This suggests that vendors positioning AI as a plug-and-play efficiency layer may be underselling the change management burden. For support leaders evaluating similar tools, the real ROI calculation should account for training investment and the organisational readiness required to shift agent workflows, not merely the headline efficiency percentage.
BBVA Cuts Customer Inquiry Response Times by More Than 15% in Italy and Germany Thanks to AI BBVA
BBVA Slashes Customer Response Times by 15% in Europe Using Generative AI FF News