BBVA: 3 hours saved per employee per week, 20,000+ GPTs, and a ChatGPT rollout to 120,000 people

• How BBVA is scaling AI from pilot to practice across the org (метрики, цитаты Браво и Альфаро, Перу 7,5 → 1 мин) — OpenAI (customer story), 2025-11-06
• BBVA and OpenAI collaborate to transform global banking (120 000 сотрудников, 25 стран, цитаты Торреса Вилы и Альтмана) — OpenAI, 2025-12-12
• BBVA and OpenAI seal a strategic alliance to redefine banking with AI (детали «The Eight», 80% ежедневного использования, демо в Италии и Германии) — BBVA (пресс-релиз), 2025-12

Background

BBVA is a global bank founded in 1857, operating across Europe, Mexico, South America, Türkiye, and the U.S., serving tens of millions of customers. For decades the bank has cultivated a reputation as a technology pioneer — chairman Carlos Torres Vila invokes that background directly: 'We were pioneers in the digital and mobile transformation, and we are now entering the AI era with even greater ambition.'

The industry context matters: banks worldwide find themselves in a dual position toward generative AI. On one hand — the strictest data and model requirements of any industry; on the other — a bank's work consists almost entirely of texts, documents, and communications, exactly the material language models handle best. Whoever first learns to combine regulatory discipline with mass AI usage gains a structural cost advantage — and the BBVA case reads as a bid for precisely that position.

The bank began working with OpenAI in May 2024, rolling out 3,300 ChatGPT Enterprise accounts, then quickly expanding access to 11,000 employees. Antonio Bravo, the bank's Global Head of Data & AI, describes the pace: 'We deployed initially to 3,000 employees, then very quickly jumped to 11,000 employees.' In December 2025, BBVA and OpenAI announced a multi-year program to bring ChatGPT Enterprise to all 120,000 employees across 25 countries — a tenfold expansion and one of the largest generative AI deployments in financial services.

Crucially, the deployment is part of an explicitly formulated strategy rather than a set of experiments. 'We've been having a lot of debates on how to make AI part of our business strategy, not a tech effort that sits on the side,' Bravo says. Those debates produced the strategic roadmap 'The Eight': redesigning the bank end-to-end — from customer experience and relationship managers to risk analysis, operations, software development, and employee productivity — developed in direct cooperation with OpenAI's product and research teams (BBVA, OpenAI). This case is about what scaling AI looks like in a heavily regulated industry when it is driven through trust and governance rather than bans.

Problem

A bank is a heavily regulated environment where the main risk of mass AI use is shadow AI: employees take work data to public chatbots, and no ban stops that — it merely removes it from oversight. Elena Alfaro, BBVA's Head of Global AI Adoption, states the alternative plainly: 'Instead of shadow AI… we gave them a platform that was safe so they could start experimenting.'

The second problem is political, and typical of large organizations: AI easily gets stuck as 'a tech effort that sits on the side', with no mandate to change processes and no board attention. BBVA deliberately framed the question differently — how to embed AI into business strategy and the daily work of every team, from legal and risk to customer service and marketing.

The third problem is scale and regulatory heterogeneity: 120,000 employees in 25 countries means different regulators, languages, data-protection jurisdictions, and levels of digital maturity. Rolling a tool out 'by decree' onto such a structure is impossible: without a built-out model of trust, training, and guardrails the bank would get either sabotage or unmanageable risk. That is why the case's key fork was the sequence: governance and culture first, scale second.

Solution

BBVA built the program on three pillars — trust, governance, and structured learning.

Trust: 'We created this atmosphere of being in a safe place to learn and to use AI,' Alfaro describes. Rather than pushing experimentation underground, the bank legalized it inside a protected perimeter: a corporate platform with guardrails where a work task can be brought without creating leak risk. Governance: security, legal, and compliance were program partners from day one — in the bank's own framing, that is what 'enabled scale, not slowing it'. Learning: dedicated training covered 250 senior leaders, including the CEO and the chairman — an organizational signal no memo can replace.

Then bottom-up energy took over: the teams closest to the work began assembling their own tools. Over 20,000 custom GPTs have been created across the bank, with around 4,000 used frequently by teams. A telling local example is Peru: an internally built assistant used by more than 3,000 employees cut query handling time from ~7.5 minutes to ~1 minute — roughly 80%. Engagement became self-sustaining: in Bravo's words, 'once you start using it, it's very sticky and… you feel it helps you a lot'; the OpenAI story cites the employees' inside joke that if ChatGPT were taken away, they'd 'have to leave'.

The bank itself formulates five managerial lessons of the program (in the OpenAI story): leadership sets the tone — 250 leaders learned the tool hands-on; governance is built as a foundation, not an afterthought; shadow AI turns into safe AI through secure environments 'always with humans in the loop'; scale early enough to see the signal — 'starting at meaningful scale surfaced real evidence quickly'; and above all, empower the front lines: the thousands of GPTs were created by the teams closest to the work, and those tools are now reused across the bank.

For customers, the bank launched Blue, a virtual assistant built on OpenAI models — it helps manage cards and accounts and answers everyday questions in natural language. The December 2025 agreement takes the program to the next level: the rollout to all 120,000 employees includes security and privacy controls, access to OpenAI's latest models, and tools for building internal agents connected to the bank's systems; a dedicated BBVA team works directly with OpenAI's product, research, and technology success teams. 'The Eight' roadmap spans eight tracks — from a conversational assistant for customers and relationship-manager tools to risk analysis, software development, and digital 'alter egos' for employee workflows; the bank has already demonstrated ChatGPT-integrated apps for its digital banks in Italy and Germany and is exploring scenarios where customers interact with the bank directly from ChatGPT (BBVA).

Result

Per figures published by OpenAI in November 2025: about 3 hours saved per employee per week, 83% weekly active usage, efficiency improvements of up to 80%+ in tests of specific workflows, and 20,000+ custom GPTs created (around 4,000 in frequent use). December materials cite an even stronger engagement metric: 80% of users access the assistant daily. It was on the back of these results that the decision to roll out to all 120,000 employees was made — and Sam Altman publicly calls BBVA 'a strong example of how a large financial institution can adopt AI with real ambition and speed'.

Frames to keep in mind. All metrics are BBVA's internal estimates published by OpenAI and the bank itself; there is no independent audit, and both parties benefit from a positive picture. 'Up to 80%+' refers to tests of specific workflows, not the whole operating model; '3 hours a week' is users' self-assessment on routine tasks, not a time study. The Peru example (7.5 min → 1 min) is one assistant in one country. Finally, '83% weekly active' and '80% daily' are metrics from different periods and methodologies and should not be glued together.

In our view, this case has two genuinely rare traits. The first is the managerial honesty of the sequence: 3,300 → 11,000 → 120,000 with interim metrics published before each next step; that makes the tenfold expansion defensible to the board and the regulator, and the case reproducible as a method (unlike irreproducible 'success stories'). The second is the pyramid of 20,000 GPTs created versus 4,000 frequently used: the bank itself publishes a 1:5 tool survival ratio, and that is the case's most useful number for planning your own rollout — impact should be computed from the active core, not the gross count of what was created.

A note of caution to close: 'The Eight' and the 'AI-native bank' are a declaration of direction, not a result; it will be testable through customer-experience and operating-cost metrics years from now, and we would not project today's productivity figures onto tomorrow's transformation promises.

Lessons learned

1. The answer to shadow AI is not a ban but a safe alternative: BBVA legalized experimentation inside the bank's perimeter and got 83% weekly active usage instead of leaks to public bots.
2. Compliance, security, and legal as partners from day one is what allowed the bank to scale fast — not what slowed it down.
3. Training 250 senior leaders including the CEO and chairman signals more than any memo: the tone is set from the top.
4. The '20,000 created → 4,000 frequently used' pyramid is a normal enterprise GPT funnel; plan impact from the active core, not the headline count.
5. Staging 3,300 → 11,000 → 120,000 with published interim metrics makes each next step defensible to the board and the regulator.
6. 'Up to 80% efficiency' in tests of specific workflows is not 80% across the bank: read the metric's frame before porting it into your own business case.
7. Local wins scale a program better than directives: the Peru assistant (7.5 min → 1 min for 3,000+ employees) became the internal proof of value the whole organization cites.

Frequently asked questions