Across APAC, CIOs and CMOs are reaching a breaking point with their marketing and customer-engagement technology landscapes. Years of adding point solutions, integrating new channels, and meeting compliance requirements have created the same outcome everywhere.
These efforts have led to technology environments that are sprawling, expensive, brittle, and difficult for anyone to fully control.
HCLSoftware’s EVP and portfolio general manager for business and industry solutions, Rajesh Iyer, told iTnews Asia, “Everybody’s tired of the Frankenstack, or some people call it the Martech jungle, that large organisations have assembled over the years: multiple data repositories, sources of truth, Martech systems, pipelines, and multiple agencies producing creative work, all leading to creative drift.”
“What you intended in the first brief is not what comes out in the final edit; you wanted an apple and got an orange. This fragmentation is a real problem,” Iyer added.
Executives no longer want “more AI” but aim for simplification. The need is to have architecture that reduces fragmentation, not another tool that adds to it.
APAC scale and complexity strain legacy marketing stacks
Enterprises across APAC engage audiences in millions, operate markets with different languages, cultural norms, regulatory rules, and consent frameworks. They manage massive operational data stores, third-party audience lists, and behavioural data streams from apps, branches, call centres, and websites.
According to Iyer, traditional AI helps with segmentation, propensity scoring, and next-best-action modelling, but it remains procedural and manual.
Teams use AI to generate lists, pick high-propensity customers, trigger messages, and run campaigns, but every wave requires humans to repeat the same steps, said Iyer.
Campaign execution improves incrementally but does not change structurally, he added.
Every year, a company adds a new tool to solve a specific problem. A CDP (customer data platform) for identity, a DSP (demand-side platform) for ads, a real-time decision engine, an email template system, a creative routing tool, and yet another platform for push or SMS.
Gradually, these setups turn into a web of loosely connected systems with overlapping functions, unused features, and significant operational waste.
Iyer mentioned that CIOs and CMOs across APAC put it, “Please don’t talk about agentic AI anymore. Show us outcomes and simplification.”
CIOs and CMOs across APAC are trying to simplify tech stacks. At the same time, they don’t want disruption. They expect what already works to keep working, and are not willing to introduce new risks around compliance or data leakage.
- Rajesh Iyer, EVP & Portfolio General Manager, Business & Industry Solutions at HCLSoftware
He added that the beauty of agentic AI is that you can have either goal.
Agentic AI becomes the new integration layer
Currently, CIOs fear that adding agentic AI might create yet another silo.
However, agentic AI, when architected correctly, does the opposite. It operates as the unifying system-of-systems that sits across existing tools and makes them work together.
Iyer mentioned that instead of being another standalone component in an already crowded stack, it operates as the integration layer that pulls data from different systems, stitches information dynamically, and applies the enterprise’s privacy, compliance, and permission rules consistently.
It generates segments automatically, executes and optimises campaigns, retargets customers in real time, and maintains always-on, multi-wave engagement cycles without requiring teams to manually coordinate every step, he added.
However, Iyer cautions that for agentic AI to function this way, strong guardrails are essential.
Well-designed agents adhere to constraints, including avoiding bias beyond approved demographic or psychographic parameters, never accessing data that a human operator would not be permitted to see, and staying within established governance policies around offers and limits.
Iyer said they remain protected against adversarial threats and provide explainability by logging the decisions they make, the reasoning behind them, and the data used.
A human override or kill switch ensures operators can halt activity instantly if needed, he added.
These safeguards allow agentic AI to operate safely within large, complex enterprises without compromising security or compliance.
Such architecture gives a unifying layer that reduces complexity and brings coherence.
All this requires enterprises to rethink the technical foundations of marketing systems to deliver real business outcomes instead of incremental gains.
Reset marketing tech foundation
Companies already sit on large pools of customer data, but none of it matters without a clean, unified master record. ID resolution and accurate customer profiles form the base layer; without them, every downstream action fails.
According to Iyer, the next requirement is real-time behavioural data.
Static attributes don’t reveal intent, but signals, including app usage, branch visits, or website research, do, and marketing only works when the system can act on those signals within minutes or hours, not days, Iyer said.
Scale adds another layer of complexity.
In markets like India, China, and Indonesia, campaigns need to reach millions while staying inside strict privacy and compliance regimes.
Data sovereignty is now a hard constraint, with enterprises demanding control over where data lives and which cloud, or on-prem platform, runs their workloads.
Iyer said HCLSoftware has designed products to operate across any cloud or on-prem environment so customers can keep data exactly where they want.
He added that the toughest challenge is proving the impact.
CMOs fund these systems, yet sales leaders only care about measurable outcomes, say, for example, how many customers were acquired, and how clearly the marketing spend connects to those results.
The next generation of marketing technology must close that loop and show causal ROI if it is to move beyond incremental gains and deliver real business value.
Scaling demands skills, compliance, and proof of real business impact
Scaling agentic AI across APAC exposes two persistent barriers: skills and regulatory fragmentation.
Iyer said, even in India, with one of the world’s largest IT talent pools, the expertise required to deploy agentic AI remains thin.
Teams lack practical experience in prompt engineering, retrieval-augmented generation, agent orchestration, and the disciplined, compliant use of AI in live environments, he added.
People are still learning how to operationalise multi-agent workflows while maintaining the human oversight that prevents AI from becoming a “bigger gun to shoot yourself in the foot with.”
Secondly, CIOs need a deliberate adoption roadmap.
Begin with high-confidence workflows, test in controlled pilots, validate guardrails, and then scale with internal capability rather than defaulting to vendor dependency.
Iyer added that regulatory diversity adds another layer of complexity.
APAC’s data protection regimes differ widely, and even mature frameworks like GDPR leave room for interpretation.
He said, “Enterprises cannot wait for perfect clarity from regulators. They must establish their own augmented internal guidelines that adhere to national laws while prioritising customer privacy in practice.”
Iyer argues that the real test is whether customers perceive interactions as respectful and empathetic.
That requires cultural localisation; what resonates in Singapore may not resonate in Indonesia or Malaysia, and messages that work in markets with mature financial systems do not translate directly to places where basic credit instruments are still emerging.
Additionally, measuring impact requires the same discipline.
According to Iyer, metrics like the number of autonomous campaigns do not matter.
Track the indicators the business already uses, including new-to-bank customer acquisition, activation and conversion rates, cost of acquisition, cycle-time reduction, and the shift in workload from humans to agents.
The task is to show what changed before and after deploying AI over 30, 60, and 90-day cycles, Iyer said.
It enhances the pace of learning, allowing teams to adapt in days instead of months, but the fundamentals remain unchanged, he added.
Agentic AI is not about adding more technology but about removing complexity and creating an automated architecture that delivers measurable outcomes at enterprise scale.
