Greg Jarvis, Head of Customer Success at Connect SA.
South Africa’s customer experience (CX) leaders are navigating a period of rapid change, driven by the promise and pitfalls of artificial intelligence (AI).
In an effort to help gauge business readiness in the sector, unpack the challenges, and explore how leading organisations are unlocking the true potential of AI, Connect hosted an executive roundtable discussion, bringing together senior leaders from across industries, with Dion Millson: Head of AI Strategy, and Greg Jarvis: Global Head of Customer Success at Connect, leading the robust discussions.
The open dialogue format focused on the evolving role AI is playing in CX and contact centre operations, and the strategies shaping the next wave of business outcomes to increase value.
Participants highlighted the market’s tendency to swing between hype and reality. High-profile AI deployments, such as Klarna’s claim of handling 2.3 million conversations with generative AI – equivalent to 700 agents – have shaken up the industry.
Value over hype. (Image: Connect)
Yet, these stories are often tempered by the reality that 70–85% of AI projects fail to reach their goals or are scrapped before full implementation, with a 42% abandoned rate among those implementing their own initiatives in 2025. This was, however, down from 70% in 2024.
Furthermore, organisations typically scrapped 46% of proofs of concept (POCs) before production deployment, and only about 30% of projects moved past staging to full implementation. Klarna has also walked back a lot of the comments originally made, and is actually re-employing agents to handle tasks that require a human touch.
Based on this reality, the panel stressed the importance of navigating this gap with caution and realism. Unsurprisingly, the boardroom conversation around AI investment is increasingly shifting from technology features to tangible business value.
However, despite the potential pitfalls as businesses work through the learning curve, the panel confirmed there are many companies achieving great things with AI, and the sector is rapidly gaining maturity.
CX leaders are already less concerned with what platforms like AWS or Genesys can offer in terms of features, and are more focused on how AI can deliver measurable outcomes a year down the line.
At the heart of the value extraction lies data. The participants all agreed that while AI use cases in business are becoming clearer, the real opportunity lies in unlocking value from data, especially the unstructured data sitting in contact centres.
The industry leaders agreed that successful AI projects start with understanding and freeing up data from multiple sources, including voice recordings, not just plugging in tools.
The ability to segment customer bases and tailor contact strategies, such as nudging good payers via digital channels, offering call-backs to low LSM customers, or saving customers time with a choice of channels and automated prompts they can accelerate processes like insurance claims, were cited as key differentiators.
Despite advances in automation, there was consensus that the human element remains vital.
In this hybrid model, AI is best deployed where it makes sense, but complex, emotive, or high-risk interactions, such as healthcare consultations or emergency calls, still require human agents.
Trust emerged as a major barrier to adoption, and building it is seen as essential for broader uptake of AI-driven services from wary consumers, particularly when it comes to sensitive engagements like a visit to the doctor.
In this context, AI can streamline transactional processes, but the need for human connection and context-specific solutions is paramount. South Africa’s unique healthcare history, with its prevalence of HIV and TB, will also shape how digital tools are adopted.
These hybrid models extend to other early-adopter industries, like financial services. Within banking, the risk landscape includes regulatory compliance, data privacy, and the need for accent neutrality in customer interactions.
As such, banks are cautiously optimistic, experimenting with POCs, but remain wary of reputational damage from failed deployments.
South Africa is also a major hub for digital-led innovation in short-term insurance. In this sector, digital insurers are leading the way in embracing AI-led automation, but traditional brokers still play a crucial role for clients who value personal relationships. The challenge is balancing automation with the need for empathy and care in claims and underwriting.
Seamless customer experience (CX) at the speed of speech.
Bring human level flow to your contact centre with Connect.
AI adoption strategies for success
When discussing strategies that deliver a return on investment (ROI) while meeting customer experience (CX) requirements, the panellists focused on the four core areas:
1. Opti-channel design.
The speakers advocated for an opti-channel approach that meets customers in their channel of choice, but also guides them to the channel that best serves their needs. This entails leveraging voice, chat, WhatsApp and e-mail, and shifting seamlessly between them to optimise the experience
Using AI to transcribe, analyse and categorise customer interactions was highlighted as a key enabler for organisations, helping to identify key topics, measure sentiment and pinpoint exceptions. These insights inform process redesigns that drive efficiency and improve customer satisfaction.
3. Collaboration and change management.
When discussing AI adoption and usage, the message was clear: Internal collaboration is critical. The biggest stumbling block is often people, not technology. Aligning incentives, fostering a culture of experimentation, and managing stakeholder expectations are essential for successful transformation.
4. Governance and risk mitigation.
Participants emphasised the need for strong governance, especially for high-impact or risky AI use cases. Quick wins are important to demonstrate value, but must be balanced with careful selection and containment of risks.
Next steps for CX leaders
To successfully navigate the transformative impact of AI on CX within the contact centre, the roundtable participants asserted that CX leaders must prioritise several strategic steps.
The foundational action is to focus on data, which involves thoroughly understanding the organisation’s current data landscape to identify specific, high-impact AI use cases that directly align with business KPIs and promise a strong ROI.
Simultaneously, leaders must strive to balance automation and empathy, strategically deploying AI for transactional or repetitive tasks where it adds clear value, while deliberately retaining skilled human agents for handling complex, sensitive, or high-stakes interactions that require genuine emotional intelligence and nuanced problem-solving.
Embracing an agile mindset is also critical; leaders should iterate and collaborate, adopting a fail-fast approach to experimentation, but they must ensure the organisational culture and incentive structures actively support this continuous learning and collaborative refinement.
As AI systems become more pervasive, organisations must proactively invest to build trust with both customers and employees by emphasising transparency in AI operations, ensuring the reliability of their solutions, and always providing context-appropriate support.
Finally, given the rapidly evolving legal landscape, it is essential to prepare for evolving regulations, stay ahead of anticipated changes and embed robust data privacy and security measures into every AI initiative from the outset.
As agentic AI becomes better and better at reasoning, more opportunities are emerging to create bespoke solutions that can start to solve increasingly more sophisticated customer queries.
However, accessing these capabilities requires more than a box-drop AI solution. In closing, the Connect team explained that when Elerian AI first started developing the technology that now powers Connect’s intelligent, data-driven CX solutions for the evolving experience economy, the team thought they were going to build a product.
However, they very quickly learned that the saturated AI-driven CX solution market doesn’t need another product. It needed people who cared about the outcome, which is how the team works with clients to deliver the outcomes and experiences they want for their customers.
Delivering on this CX strategy boils down to understanding a customer journey and how to leverage the best technologies to improve it. This is a vital distinction in approach as local CX leaders chart a pragmatic path towards an AI-enabled future.
By focusing on value, data and the human touch, the roundtable participants affirmed that CX leaders can turn challenges into opportunities and set the stage for the next evolution in their CX.
While the journey is ongoing, the insights shared in this roundtable offer a roadmap for others looking to deliver real business outcomes with AI.
Frequently asked questions. (Image: Connect)
Frequently asked questions
What is the current success rate of AI projects in customer experience and contact centres?
Approximately 70%–85% of AI projects still fail to reach their goals or are abandoned entirely. However, the abandonment rate for organisations building their own AI initiatives has improved from 70% in 2024 to 42% in 2025. Additionally, around 46% of proofs of concept (POCs) are scrapped before production, with only about 30% of projects moving to full implementation.
How are South African CX leaders balancing AI automation with the human touch?
South African CX leaders are adopting a hybrid model: deploying AI for transactional, repetitive, or low-complexity tasks (e.g., claims processing, nudges to good payers via digital channels) while deliberately retaining skilled human agents for complex, emotive, high-risk, or sensitive interactions such as healthcare consultations, emergency calls, or situations requiring empathy and cultural nuance.
What are the four key strategies CX leaders should focus on for successful AI adoption in contact centres?
According to industry leaders at the Connect roundtable, the four core pillars for ROI-driven AI success are:
- Opti-channel design – meeting customers on their preferred channel while intelligently guiding them to the optimal channel (voice, chat, WhatsApp, etc).
- Data-driven insights – transcribing, analysing and categorising interactions to uncover topics, sentiment and exceptions that drive process redesign.
- Collaboration and change management – aligning people, incentives, and culture; the biggest barrier is usually internal resistance, not technology.
- Governance and risk mitigation – strong oversight, especially for high-impact use cases, balancing quick wins with reputational and regulatory risk containment.
