AI agents will serve you now

Due to price sensitivity, consum­ers are expected to become more disloyal to brands in 2025. However, research firm Forrester suggests they will likely join more loyalty programmes to secure value, including personal­ised services, wherever possible.

Given consumers’ ever-changing requirements and the continued lack of manpower, how can organisations deliver highly personalised services to build lasting brand loyalty? Enter artificial intelligence (AI) agents or agentic AI.

Regarded as the next phase of gen­erative AI, AI agents can autonomous­ly perceive their environment, make decisions and take action to achieve specific goals. “Till now, most gen­erative AI applications were able to generate content or power chatbot-style applications. But they were not able to directly take action. AI agents bridge this act between generation and action, allowing generative AI applications to plan, reason, collaborate with other specialised agents, and use external tools such as functions and application programming interfaces to convert in­tention into action,” says Leslie Joseph, Forrester’s principal analyst.

Despite its capabilities, AI agents are meant to work alongside — instead of replacing — humans to drive customer success. “With real-time adaptability and deep integration into business workflows, AI agents can instantly scale workforces, boosting productivity and allowing human employees to focus on strategic, high-value tasks, driving greater customer success at every touchpoint,” says Gavin Barfield, vice president and chief technology officer for Solutions at Salesforce Asean.

AI agents in customer service

The clear value of AI agents lies in au­tomating low-value, routine tasks to boost operational efficiency. However, using AI agents solely for back-end op­erations offers limited benefits. A McK­insey study reveals that 75% of AI’s value comes from areas such as custom­er operations, marketing and sales — front-office functions where AI agents can directly engage customers and drive measurable business outcomes.

See also: Safeguarding trust in the digital age: A blueprint for financial institutions

Barfield recommends using AI agents to handle basic customer enquiries — such as tracking shipments or providing known product information — and hand off more complex cases to human agents when necessary. This improves overall efficiency and elevates the customer experience by allowing human agents to provide a higher level of personalisation.

AI agents, he adds, can help with lead qualification to automate the initial stages of customer engagement. “This includes engaging leads via email and qualifying them based on predefined criteria, freeing up sales teams to focus on closing deals with qualified leads and nurturing important prospects,” he says.

Out-of-box solutions

See also: Conducting secure data movements in the cloud symphony

The benefits of AI agents are believed to be far-reaching across any indus­try and company size. “Industries where AI agents shine the most are ones that experience a high volume of customer interactions, repetitive tasks and significant seasonal varia­tions, such as retail, financial servic­es and tech. [However, AI agents not only help large] enterprises with the sheer scale and complexity of their customer interactions, [but can also aid] small businesses that often have tighter resources and budgets,” says Maureen Chong, regional vice presi­dent for Asia at Zendesk.

Tech companies like Salesforce and Zendesk offer out-of-the-box agentic AI solutions to enable businesses of all sizes to embrace AI agents.

Companies that opt for a do-it-yourself approach to AI face higher costs due to the need for in-house expertise, model training and infra­structure development. In contrast, businesses that adopt out-of-the-box solutions like Salesforce’s Agentforce benefit from lower upfront costs. These AI agents come pre-built with low-code tools for easy customisation and rapid deployment, allowing for faster returns in terms of efficiency and operational cost reductions.



Gavin Barfield, vice president and chief technology officer, Solutions, Salesforce Asean

Wiley, for example, saw a 40% increase in case resolution after using Agentforce to streamline its customer service operations. The global pub­lishing company deployed AI agents to handle routine queries such as account access and payment issues, which had traditionally consumed a large portion of its customer service team’s time. This allowed human agents to focus on more complex, high-value tasks, significantly boosting productivity during peak periods like the back-to-school season.

Besides designing its AI agents to be purpose-built for customer experi­ence, Zendesk offers outcome-based pricing to help businesses address the concern of realising ROI from AI agent deployment. “Our outcome-based pricing for AI agents directly ties costs to tangible value by ensuring businesses only pay for queries fully resolved autonomously by AI agents. This pricing model makes AI more accessible to businesses of all sizes and budgets and ensures businesses only pay for what works,” says Chong.

This is exemplified by the cosmet­ics brand Lush, which developed a customer AI agent on Zendesk. The AI agent managed most repetitive queries, including sales and discounts, donations, order issues and discontin­ued products. By requesting customer information upfront and tagging in­coming tickets, the AI agent provided human agents with the context needed to resolve issues more quickly.

To stay ahead of the latest tech trends, click here for DigitalEdge Section

“Lush realised a 369% ROI upon its implementation of Zendesk, re­covering its initial investment in the software in less than one year. Addi­tionally, Lush was able to improve service agent productivity by 17% and manager productivity by 30%. These productivity improvements have translated to over US$434,000 ($584,272) in annual cost savings from avoided headcount,” adds Chong.

Scaling up the use of AI agents

According to research firm IDC, business leaders in Asia Pacific will demand an 80% success rate on generative AI initiatives (including AI agents) by 2027. This calls for organisations to move beyond pilot projects and embed AI into core operations. 

To do so, organisations must first identify use cases that can deliver the most returns for the business. “A good starting point is identifying repetitive, low-value tasks and workflows for AI agents to automate, allowing the role of human agents to evolve into a su­pervisory role,” says Chong.

Joseph adds that organisations need to integrate AI agents into their broader automation strategy.

Relying solely on agentic AI without considering the overall automation landscape can lead to fragmented processes, broken workflows, and unexpected customer outcomes. Effective integration re­quires careful planning to ensure AI agents work seamlessly with existing technologies and enhance, rather than disrupt, business operations. Map out how AI agents will interact with other systems within existing processes and data.



Leslie Joseph, principal analyst, Forrester

On the technical side, unifying data and ensuring data accessibility is key. “AI agents need trusted data and a complete and integrated platform to generate meaningful insights. This is a challenge as many businesses have fragmented data across multiple sys­tems. On average, organisations use 1,061 applications, with over 71% disconnected. This means structured data — such as transaction records, inventory and customer profiles — are often fragmented across systems, making it difficult for AI to get a unified view of data to deliver accu­rate, contextualised, and actionable outputs,” adds Barfield. Moreover, a large portion of unstructured data – such as documents, PDFs, audio files and emails — that may contain valuable insights to inform generative AI (and AI agents) outputs remains largely untapped.

Platforms like Salesforce’s Data Cloud can help organisations unify their structured and unstructured data. “By enabling AI agents to gain access to a comprehensive view of both a customer and the rich knowledge base of the business, AI agents can make more intelligent and informed decisions as well as act with greater accuracy and effectiveness,” says Barfield.

Data Cloud, he adds, enables sub-second real-time data processing, allowing AI agents to act on up-to-the-minute information. “Whether resolv­ing a customer query or automating a business process, this capability ensures that AI-driven actions are timely, relevant, and efficient.”

As organisations increasingly use multiple clouds for cyber resilience or regulatory compliance, they should also address multi-cloud connectivity needs.

Barfield says: “Instead of dupli­cating data across different cloud systems, businesses can leverage platforms that offer open connectivity. For example, Salesforce’s Data Cloud provides Zero Copy capabilities, en­abling businesses to connect their existing data infrastructure without replicating data, thereby maximising the value of existing data investments while expanding AI capabilities.”

Data guardrails and AI ethics

Since AI agents interact with data, including sensitive information, there is a need to apply strong guardrails to protect data from unauthorised ac­cess. “As with all AI, organisations must apply a robust set of guard­rails and security protocols, includ­ing encryption, access controls, and regular audits to detect and address vulnerabilities, identify prompt in­jection-based attacks, and avoid cas­cade failures. They must also estab­lish clear data governance policies that dictate how AI agents collect, store, and use data,” says Forrester.

To help organisations address this, Salesforce’s Data Cloud includes policy-based governance, custom­er-managed encryption keys, and AI tagging and classification features. These tools ensure that data is ap­propriately managed, encrypted, and classified, limiting access to only authorised users, whether hu­man or agent. Barfield says this helps businesses comply with strict data regulations while ensuring custom­er data remains secure across all platforms.

AI ethics is another area organisa­tions should address before deploy­ing AI agents. Zendesk’s CX Trends Report 2024 reveals that roughly one in two consumers across Asia Pacific are concerned about bias or discrimination in AI algorithms and decision-making. Therefore, organisations should adopt AI agents designed with ethical AI principles in mind to maintain customer trust.

Chong emphasises that AI solu­tions must be designed with privacy, security, and compliance at their core. AI providers should do their part by:

• Building in certain privacy func­tions, such as anonymising data used for training, limiting live chat data usage, offering opt-outs, and implementing robust data reten­tion and deletion policies.
• Being transparent and upfront about how their technology works, the data it collects, and how it is used. Providing opt-outs for businesses and their end cus­tomers fosters trust, and when it comes to quality, sharing the confidence levels of AI predic­tions helps users make informed decisions.
• Ensuring their AI solutions deliv­er accurate and reliable respons­es as well as enhance customer experience instead of replacing human interaction altogether.

Reducing bias is also important, and it can be achieved through diverse development teams and input from subject matter experts. While AI agents are designed to operate autonomously, human oversight is essential to ensure that they are trained carefully, potential bias is detected early, and continuous improvements are made. Even with out-of-the-box solutions, human agents should be responsible for ensuring the AI solution feels custom-built. This is achieved by training it on the latest data, testing response accuracy, and monitoring for issues.



Maureen Chong, regional vice presi­dent for Asia,  Zendesk.

Delivering highly personalised services is resource-intensive but not unattainable. By leveraging AI agents to handle routine tasks, organisations can build a limitless workforce, free­ing human employees to focus on high-value activities that drive cus­tomer success.