The Central Bank of the Philippines' manual complaints handling process had multiple limitations that caused protracted complaint resolution times for consumers and administrative burdens for consumer protection staff.

On the consumer side, traditional channels such as written letters and walk-in kiosks were outdated in the context of rising mobile penetration in the Philippines, the texting capital of the world. Additionally, channels such as email required a level of literacy and digital infrastructure out of reach for low-income, rural, and historically-marginalised Filipinos. The overall recourse process therefore had an urban, male, and high-income bias, resulting in the vast majority of complaints originating within the Metro Manila area, while leaving customers outside the capital with fewer to no options in cases of financial misconduct.

On the Central Bank's side, the laborious nature of case management was a binding constraint on the ability to process complaints scaling 87% year-on-year during the pilot-phase and 300% most recently (2020-21). Delays impaired timely remediation, with the risk of discouraging usage of the consumer protection mechanism. Additionally, human data entry was prone to errors and omissions, resulting in an inconsistent consumer protection database for informed policy-making. At the core of the problem was the question:

How does a regulator discover and resolve consumer protection problems if it is only receiving a small fraction of total complaints?

Proto deployed the Central Bank's Online Buddy (BOB) chatbot as a replacement for the manual process, enabling consumers to seek financial misconduct recourse regardless of geography, gender, income status or education level. From the consumer perspective, the chatbot is widely-available via Messenger, SMS, Webchat and Viber.

The chatbot is capable of understanding emotional and typo-heavy complaints in Tagalog, English, and Taglish (mixed Tagalog and English) across 776 of financial operators and 24 product & service categories. Initially deployed with an average chat-level intent classification accuracy of 92.84% for local languages, the chatbot has gathered training data from over 12,200 consumer chats per quarter in order to refine the NLP models for continuous dialogue and accuracy improvement.

All complaints are automatically categorised, converted to cases within Proto's Tickets dashboard, and disseminated to the financial operators for reply within 7 days. All ongoing communication between the consumers and financial operators is monitored by the Central Bank. The consumer protection staff are thus repurposed for higher-order functions such as consumer case investigation, overdue case follow-up, and policy development based upon the consistent data set gathered by the automated system. Additionally, the data offers efficient insight into potential market misconduct, with gender disaggregation, which was not provided by the previous manual system. Crucially, the data also includes insights via the SMS channel from rural and historically-marginalised populations that may not yet own smartphones.

From an innovator's perspective, working directly with regulators is not an intuitive approach to scaling a certain technology. However, Proto's market entry strategy focuses on first deploying its AICX solutions with industry regulators. In the emerging world, this approach provides multiple, unparalleled benefits.

First, localised Al technology has inherent capacity for good by including and protecting underserved and marginalized people who face historical access barriers. The regulators are empowered by consumer protection mandates and legal frameworks, but they have resource constraints limiting their ability to keep pace with disruptive technologies that pose potential risks for consumers.

Second, and most essential, the regulator is effectively a "data hub" by way of its mandate to gather and monitor consumer protection communication in its industry. For example, the Central Bank gathers structured local language data from 776 financial operators. For an Al innovator, this training data from a single source is a highly efficient pathway to achieving a continuous machine learning loop or network effect at massive scale. The result of this approach is superior, localised NLP models for the industry domain that lead by example and set a new standard for consumer engagement for the private sector to emulate.

The impact of this Proto AICX deployment is best captured with the story of Kitty Elicay, a financial consumer in Manila who suffered an 8 month credit card refund problem during the COVID-19 pandemic – she used the Central Bank's chatbot to effect a monitored resolution with her financial institution that saw the funds returned within six days.

The Central Bank of the Philippines is pursuing a balanced approach to the risks of digital financial services by keeping pace with technical innovation, deploying leading NLP capability, and addressing known gaps in consumer protection. In recognition of this effort, the automated consumer protection system went on to win the Central Banking Fintech & Regtech Al Prize and contributed to Proto's award of the Inclusive Fintech 50 by Visa, MetLife, Accion and IFC.

As digitisation expands financial services into underserved population segments, regulators need to match this increased volume of first-time consumers with protection systems designed for equal scale and complexity. Reliance upon existing manual processes risks failing to comply with consumer protection legislation and abandoning financial consumers to misconduct without recourse. Failure to protect consumers from these risks also affects confidence in financial technology, which could stifle growth of indigenous financial providers. Consumer protection is therefore closely entwined with efforts to improve inclusion and stability.

These themes are explored further in CGAP’s Inclusive Finance Frontiers podcast episode, Can AI Revolutionize Financial Consumer Protection? featuring representatives from the BSP, Cambridge SupTech Lab, and CGAP. The conversation frames AI not just as a regulatory tool for efficiency, but as an enabler of consumer trust, faster risk detection, and multilingual engagement at national scale. These principles are reflected in Proto’s deployment with the Bangko Sentral ng Pilipinas and expanded through the Gates Foundation–funded shared service model, which unifies more than 30 Philippine government agencies into a single AI-powered redress system.

The model, implemented with Cambridge SupTech Lab, operationalizes mandates from the Internet Transactions Act by providing interoperable AI assistants and case management across regulatory bodies. It is the first national deployment under a multi-country initiative to scale inclusive grievance systems in low- and middle-income countries, with further launches underway in Africa. A core feature of the model is its cross-agency architecture—enabling coordinated intake, resolution, and analytics while reducing duplication and operational costs. The approach sets a precedent for how public infrastructure can adapt to the scale and complexity of digital consumer protection.

As CGAP’s Eric Duflos notes in the podcast, responsible use of AI in supervision demands more than automation—it requires a collaborative ecosystem of authorities aligned around inclusion, transparency, and long-term accountability. This is precisely the direction Proto’s shared service model is designed to support.

Proto is the leading generative AICX platform for local languages. Its inclusive text and voice AI assistants excel at usecases for customer experience, consumer protection, employee experience, and indoor navigation. The Proto AICX Platform is powered by large language models and the proprietary ProtoAI engine for high accuracy in underserved languages. Proto deployments feature enterprise-grade security and capabilities such as on-premise hosting, customised analytics, and a 24/7 prompt engineering service. Headquartered in Waterloo, Canada, Proto is a global team operating across Latin America, Africa, and Asia.