By continuing to visit this website, you agree to our privacy policy and use of cookies.

Agree
Article

In 2021, AI customer experience is about inclusion

5

min read

February 12, 2024

CEO Curtis Matlock shares Proto’s blueprint for inclusive customer experience automation in the year ahead. Proto is an AICX vendor for business and government contact centres in the emerging world.

The past year — more chatbots but less equality

For customer experience automation, 2020 was a year of opportunity and reflection. When lockdowns forced crowded contact centres to empty, enterprises adopted automation tools to maintain customer service – including chatbots and natural language processing engines. Within the financial services sector of the chatbot industry, growth is unphased, with a projected size of $3.39 billion by 2027, and a compound annual growth rate of 27.3%. Proto has seen from its customer touchpoints that these projections are likely outdated due to the pandemic convincing even late-adopters to perceive the benefits of automation.

While the AI customer experience industry is surging, many consumers are falling behind. Consider this contrast: vaccines will be offering citizens relief from the daily danger of COVID-19, but the longer-term societal impacts of the pandemic will continue to crush the most vulnerable. Some examples:

Put another way: while some advantaged industries, such as customer experience automation are gaining, citizens around the world are struggling. Companies like Proto have the technology to help narrow the inclusion gaps widened by the pandemic, and consumer-facing enterprises adopting automation have a role to play. Why?

AI and inclusion in action

Messaging technology is the most available channel through which consumers can communicate their questions and frustrations to receive a fair and timely chance at resolution. The application of AI to messaging offers a new opportunity for opening access to the underserved communities most affected by the pandemic and pre-existing inclusion gaps.

AI is a force multiplier for the human agents within enterprises and governments that provide critical services to consumers. The right automation solution can be applied to reduce wait times, instantly perform repetitive and laborious tasks, and transform the consumer experience across an unlimited volume of cases. Proto saw first-hand how this works for the Central Bank of the Philippines (BSP) during the pandemic:

The BSP faced an unsustainable consumer complaints volume growth of 87% year-on-year, via traditional channels that were known to favour urban consumers, such as email and walk-ins. To solve the problem of service capability and inclusion, the BSP’s chatbot was deployed via Messenger, SMS and webchat, and understands emotional and typo-heavy consumer complaints in Tagalog, English, and Taglish (mixed Tagalog and English) across 776 financial institutions nationwide. Initially deployed with a 92.84% chat-level intent classification accuracy, the solution improves itself with structured training data from 12,200 consumer chats in the last quarter alone.

A critical part of the BSP’s journey has been putting inclusion at the heart of its automation agenda.

Before the pandemic, automation procurements were typically driven by cost savings, with impact areas such as financial inclusion and consumer protection dropped to a lower-priority. Rethinking this prioritisation can unlock the most common benefits of automation, while better supporting the underserved communities that are most affected by the pandemic.

Automation is more than cost savings

Instead of procuring with the primary lens of cost savings, there is an opportunity to prioritize inclusive automation solutions. This type of technology should offer machine-understanding of indigenous and mixed languages, and the proven capability to reach bottom-of-the-pyramid consumers via feature phones and messaging apps. By adopting localised automation, enterprises have a best method to provide access to historically-marginalised populations, such as those from rural areas and ethnic language groups.

You might ask, will localised AI be more expensive? It could be if contracts are awarded to consulting firms seeing a quick opportunity to multiply their revenue with value-added services. However, these vendor options may lack the data required for local-language AI models in countries such as Ghana or the Philippines. Also, their business model may not even be designed to guarantee long-term impact outcomes. For better alignment, enterprises should look for vendors with the following four attributes:

1. Software-as-a-service.
The consultancy model typically has a high upfront cost and a deteriorating service standard over time, unless additional maintenance fees are paid. These maintenance fees are either ad-hoc (resulting in potential overruns) or locked-in (resulting in unused but billed hours). Instead, the long-term licensing from the SaaS model ensures a reliable service standard with predictable costs, feature upgrades, and software patches, which ultimately results in better customer experience and consistent inclusion outcomes.

2. Impact focus. For-profit SaaS vendors may have innovative technology, but also having a track-record of impact-focused deployments is rare. Vendors that have mastered both the tech and the impact should be prioritised.

3. Regulator clients. Vendors that also work with industry regulators appreciate national priorities, and have the insight to design solutions for prevailing inclusion pain points. These relationships can also provide vast amounts of rare training data for AI models.

4. Local presence. Even in a world familiar with online meetings from distant locations, there is no substitute for having local personnel collaborating on every component of a vendor’s business — from the go-to-market strategy to product design. Solutions will be far more impactful when vendors have local skin-in-the-game.

Where these attributes are prioritised, automation can go beyond a cost-savings initiative, and enterprises can deliver real impact from their AICX journey.

2021 — the year for inclusive AICX

Rethinking AICX to put inclusion first is not easy. A commercial bank in Rwanda would be pressed to find a localised, impact-focused SaaS vendor with competitive AI models for the Kinyarwanda language, plus a history of service to the national financial regulator.

This is because simultaneously building an impact-focused SaaS business and generating AI models for local languages, such as Tagalog and Hausa, is a tall order for the best software development teams. While there is no one perfect roadmap to follow at the frontier of a new industry, Proto has a blueprint with six fundamentals to deliver the most inclusive AICX solutions possible in the year ahead:

1. Find the data hubs.
Chatbots that understand local languages (i.e. Cebuano, the most common language of the southern Philippines) or mixed languages (i.e. Sheng, a mix of Swahili and English) need to be powered by AI models with training data from those languages. The problem is that these models are under-resourced, meaning there is a scarcity of available training data. Even rarer is training data specific to industry domains, such as financial services terminology for Cebuano. Typically, solving this problem is a long process where vendors go from client-to-client collecting data, and network effects are slow — if not impossible — to create. A unique shortcut around this problem is working with data hubs, organisations within industry ecosystems that already collect this data from across hundreds of operators. Regulators such as central banks, gaming commissions, and national health organisations are ideal data hubs by way of their consumer protection mandates. Proto has spent the past years building its partnership framework with regulatory authorities throughout the emerging world, and we’ll be rolling out consumer protection solutions in 2021 that harness this vast language-domain training data from entire industry ecosystems.

2. Localize the NLP team. Location-dependent AI vendors, usually based in one country in Europe or North America, can have brilliant but monolingual natural language processing engineers. These are the folks that do the hard work of structuring training data and fine-tuning the AI models, but there’s only so much they can effectively do if they cannot understand the natural language of the model. Proto knew from the get-go that recruiting NLP engineers in Vancouver who also spoke Thai, Yoruba, and Tagalog would be a non-starter. To overcome this, we’ve built recruitment partnerships with machine-learning institutes, such as the African Institute of Mathematical Sciences, where our last six NLP engineer hires (covering over 10 languages) were trained. In 2021, Proto will continue to be a remote team that attracts top talent from within the markets that we serve.

3. Don’t only chase SaaS metrics. Focusing on the strategic work of AI model building, rather than the typical SaaS metrics, is essential for building AICX that’s going to create long-term value. Startups — especially those striving to meet the expectations of venture capitalists — are compelled to measure themselves by the metrics of the past decade of enterprise SaaS. As a result, management teams can get caught in a paradoxical mission to achieve both >100% year-on-year recurring revenue growth and critical AI metrics such as decreasing rate of human-intervention and algorithm rev-up. This can be a source of friction in which a company may prioritize the SaaS-side of the business over the low-margin but essential work of AI model building. According to AI-ROI.com, this would be a mistake:

“Without the reinforcement loop generating a compounding volume of data and an increasingly powerful AI over time, that company’s product remains vulnerable to copycats and will eventually be commoditised… AI offers the opportunity to deliver the customised and specialised ROI of a services business with the scalability of software, with the ability to defend against copycats. The high start-up costs of this approach to company-building may mean you will realise smaller profits and build the company prioritising different elements than what has worked before.”

Proto has spent the past years investing in its local language AI models, and will continue to prioritise the AI metrics that drive long-term client loyalty and product superiority.

4. Focus on client solutions (not hype). There’s been a lot of hype around AI that can be too centred on the technology’s potential instead of its solutions to customer problems. For example, following the raise of US$102M in 2017 and $25M in government grants this past year, a once-celebrated AI leader announced its acquisition, where the founders had their share value “wiped out” and employees were “terminated and had their stock options cancelled,” according to The Globe and Mail. This disappointing outcome occurred after the leading AI company “took two years to focus on product development after initially pursuing consulting gigs.” In 2021, Proto will continue to focus its time and capital on enhancing its core product to deliver real impact for consumers.

5. Treat your communities well. It should go without saying, but questionable moves by some firms and investors during the pressures of 2020, such as the cancelation of employee stock options and pulling terms sheet, are demotivating for teams with an inclusion mission. With 93% team ownership, Proto is fully aligned with the interests of its employees and clients, and will value prospective partners that also resonate with the tech-for-good culture.

6. Take direction from impact leaders. The daily work of building, deploying and refining software can leave little time for thinking through inclusion blindspots. By partnering with leading associations and businesses in the impact world, such as the RegTech for Regulators Accelerator and Bankable Frontier Associates, Proto ensures ongoing feedback and arms-length impact analysis of our deployments. Proto also maintains its annual participation in the Alliance for Financial Inclusion events, to tap into the thinking of its global regulator membership. In 2021, we’re looking forward to building more partnerships that keep Proto’s technology in pursuit of impact outcomes.

With these fundamentals, the Proto team will be focused during the year ahead on providing enterprises and governments with the AI-powered messaging technology to include all consumers, regardless of language, geography, ethnicity or gender. We’re excited to show how AICX can deliver efficiencies and play an essential role in improving inclusion for the years to come. To learn more amount how Proto can help your business include local customers, book a product tour with our experts.

About Proto

Proto is the leading generative AI platform for local customer experience. Powered by GPT and its own proprietary AI engine, Proto's AI Customer Experience (AICX) Platform automates transactions, tickets, bookings, and other high-volume interactions for local and mixed languages across text, voice and 20+ messaging channels. Proto is globally-recognized for its inclusive AICX solutions across the government, financial, health and transport industries.

View Case Study
linkedin share button
whatsapp share button