As Southeast Asia accelerates AI adoption, tightening guardrails around agentic AI are pushing accountability higher up the agenda.
Singapore’s Prime Minister put AI risk in the spotlight during his recent National Day Rally speech, warning that AI agents will increasingly act with less human supervision and citing a real case of an agent breaching its test environment to attack a company.
The warning follows a concrete regulatory move, with the city-state’s updated Model AI Governance Framework for Agentic AI introducing safeguards to counter emerging risks and keep human users in control. Vietnam’s AI Law, ASEAN’s first binding AI legislation, took effect earlier this year, while Malaysia, Thailand, Indonesia and the Philippines are each shifting toward more enforceable frameworks of their own.
As AI-assisted development becomes mainstream, vibe-coded tools are increasingly in direct contention with those available in the marketplace. Technology leaders weighing whether to build or buy need to price in the regulatory exposure of a homegrown monitoring tool from the outset.
Accountability by design
Observability, the practice of understanding a system’s performance from the data it generates, is a build-versus-buy decision that is no longer based on engineering capacity alone. Data governance and a rapidly evolving regulatory environment now carry equal weight.
As modern enterprise infrastructure grows more complex, engineers rely on observability for deeper insight into system behavior, helping drive faster decisions and measurable gains in performance and reliability. It’s no surprise adoption is rising: 50 percent of Southeast Asian IT and engineering leaders indicate that they receive a three- to five-times return on investment from their observability platforms.
A critical benefit of observability is that it helps engineers meet rigorous regulatory and compliance standards. An established platform used by global enterprises has typically already undergone comprehensive testing, oversight, incident reporting and audit. This represents an important capability that can help users meet the standards regulators across the region are converging on.
A team that vibe-codes its own monitoring stack inherits that same governance obligation without necessarily having the scaffolding an established vendor would already have in place.
A homegrown tool can also add to the maintenance burden on an R&D organization that may already be stretched thin. Engineers who build monitoring in-house must maintain it alongside the product itself, potentially forcing teams to slow the roadmap or allow coverage to slip.
The fragmentation risk
AI coding tools put a working prototype within reach of almost any team, so new monitoring tools may emerge at the business-unit level rather than the organizational level.
A team in Jakarta might spin up its own stack, disconnected from one built in Manila or Kuala Lumpur. Each may save resources by working with a smaller data set, but the organization risks losing the cross-system correlation that makes observability valuable in the first place.
A regional bank running distributed systems across several markets, for instance, may find that fragmentation turns a routine incident into a multi-jurisdiction investigation.
Scale compounds the problem further. Established vendors have spent years accumulating the correlations and anomaly patterns that can make root-cause analysis faster. A vibe-coded tool starts from zero, while competition for the specialist talent needed to run bespoke systems only intensifies as more organizations try to build their own.
The real cost of DIY
A narrow, fast-built tool can still solve an immediate problem cost-effectively. But the case for buying deserves serious consideration, especially for large, distributed enterprises that have more to lose from fragmented incident response and face greater regulatory scrutiny over how they govern data and automated systems.
Smaller, agile companies are often assumed to be AI’s biggest beneficiaries. In observability, however, a lean startup may be less able to absorb the cost of maintaining a second platform alongside its core product, particularly as governance obligations increase.
Every established vendor has an incentive to argue against DIY, and the AI era makes building look faster and cheaper than ever. But organizations across Southeast Asia also need to account for a compliance dimension that is only growing.
As regulators demand greater accountability and governance, every tooling decision increasingly comes with obligations that leadership needs to be ready to own.
Virginia Galarón Herrera is Senior Director of Technical Success at New Relic.
Editor’s note: This article is a contributed perspective. Quantitative findings, regulatory characterizations, and assessments of technology approaches reflect the author’s cited sources and perspective. The article has been lightly edited for clarity and TNGlobal house style.
Share your perspective: TNGlobal welcomes contributed insights and expert commentary from across Asia’s technology and innovation ecosystem. Submit a contribution for editorial consideration, or explore more perspectives in TNGlobal INSIDER and conversations in the TNGlobal Q&A and Interviews archive.
Featured image: Kevin Ku on Unsplash