Agentic AI is being talked about as the subsequent main wave of synthetic intelligence, however its which means for enterprises stays to be settled. Capgemini Research Institute estimates agentic AI may unlock as a lot as US$450 billion in financial worth by 2028. But adoption remains to be restricted: solely 2% of organisations have scaled its use, and belief in AI brokers is already beginning to slip.
That stress – excessive potential however low deployment – is what Capgemini’s new analysis explores. Primarily based on an April 2025 survey of 1,500 executives at massive organisations in 14 nations, together with Singapore, the report highlights belief and oversight as vital elements in realising worth. Almost three-quarters of executives stated the advantages of human involvement in AI workflows outweigh the prices. 9 out of ten described oversight as both optimistic or no less than cost-neutral.
The message is obvious: AI brokers work greatest when paired with folks, not left on autopilot.
Early steps, sluggish progress
Roughly 1 / 4 have launched agentic AI pilots, whereas solely 14% have moved into implementation. For almost all, deployment remains to be within the starting stage. The report describes this as a widening hole between intent and readiness, now one of many predominant boundaries to capturing financial worth.
The know-how is not only theoretical – real-world functions are beginning to emerge, and one instance is a private purchasing assistant that may seek for objects based mostly on particular requests, generate product descriptions, reply questions, and place objects in a cart utilizing voice or textual content instructions. Whereas these instruments sometimes cease in need of finishing monetary transactions for safety causes, they already replicate lots of the capabilities of a human assistant.
This raises larger questions concerning the function of conventional web sites. If AI can deal with duties like looking out, evaluating, and making ready purchases, will folks nonetheless must navigate on-line shops immediately? For individuals who discover busy web sites overwhelming or troublesome to navigate, an AI-driven interface might provide a less complicated, extra accessible possibility.
Defining agentic AI
To chop by the hype, AI Information spoke with Jason Hardy, chief know-how officer for synthetic intelligence at Hitachi Vantara, about how enterprises in Asia-Pacific ought to take into consideration the know-how.
“Agentic AI is software program that may determine, act, and refine its technique by itself,” Hardy stated. “Consider it as a staff of area specialists that may be taught from expertise, coordinate duties, and function in actual time. Generative AI creates content material and is often reactive to prompts. Agentic AI might use GenAI inside it, however its job is to pursue aims and take motion in dynamic environments.”
The excellence – between producing outputs and driving outcomes – captures the which means of agentic AI for enterprise IT.
Why adoption is accelerating
Based on Hardy, adoption is being pushed by scale and complexity. “Enterprises are drowning in complexity, threat, and scale. Agentic AI is catching on as a result of it does greater than analyse. It optimises storage and capability on the fly, automates governance and compliance, anticipates failures earlier than they happen, and responds to safety threats in actual time. That shift from ‘perception’ to ‘autonomous motion’ is why adoption is accelerating,” he defined.
Capgemini’s analysis helps this. The research discovered that whereas confidence in agentic AI is uneven, early deployments are proving helpful when the know-how takes on routine however important IT duties.
The place worth is rising
Hardy pointed to IT operations because the strongest use case to this point. “Automated information classification, proactive storage optimisation, and compliance reporting save groups hours every day, whereas predictive upkeep and real-time cybersecurity responses scale back downtime and threat,” he stated.
The impression goes past effectivity. The capabilities imply techniques can detect issues earlier than they escalate, allocate sources extra successfully, and comprise safety incidents extra shortly. “Early customers are already utilizing agentic AI to remediate incidents proactively earlier than they escalate, strengthening reliability and efficiency in hybrid environments,” Hardy added.
For now, IT stays essentially the most sensible start line: its deployment presents measurable outcomes and is central to how enterprises handle each prices and threat, exhibiting the which means of agentic AI in operations.
Southeast Asia’s start line
For Southeast Asian organisations, Hardy stated the primary precedence is getting the info proper. “Agentic AI delivers worth solely when enterprise information is correctly labeled, secured, and ruled,” he defined.
Infrastructure additionally issues, which means that agentic AI requires techniques that may help multi-agent orchestration, persistent reminiscence, and dynamic useful resource allocation. With out this basis, adoption shall be restricted in scope.
Many enterprises might select to start with IT operations, the place agentic AI can pre-empt outages and optimise efficiency earlier than rolling out to wider enterprise capabilities.
Reshaping core workflows
Hardy expects agentic AI to reshape workflows in IT, provide chain administration, and customer support. “In IT operations, agentic AI can anticipate capability wants, rebalance workloads, and reallocate sources in actual time. It may additionally automate predictive upkeep, stopping {hardware} failures earlier than they happen,” he stated.
Cybersecurity is one other space of promise. “In cybersecurity, agentic AI is ready to detect anomalies, isolate affected techniques, and set off immutable backups in seconds, lowering response occasions and mitigating potential harm,” Hardy famous.
The capabilities should not restricted to proof-of-concept trials. Early deployments already present how agentic AI can strengthen reliability and resilience in hybrid environments.
Abilities and management
Adoption may even require new human expertise. “Agentic AI will shift the human function from execution to oversight and orchestration,” Hardy stated. Leaders might want to set boundaries and monitor autonomous techniques, making certain they keep in moral and organisational limits.
For managers, the change means much less concentrate on administrative duties and extra on mentoring, innovation, and technique. HR groups might want to construct governance expertise like auditing readiness and create new constructions for integrating agentic AI successfully.
The workforce impression shall be uneven. The World Financial Discussion board predicts that AI may create 11 million jobs in Southeast Asia by 2030 and displace 9 million. Ladies and Gen Z are anticipated to face the sharpest disruptions, with greater than 70% of girls and as much as 76% of youthful staff in roles susceptible to AI.
This highlights the urgency of reskilling, and main investments are already underway, with Microsoft committing $1.7 billion in Indonesia and rolling out coaching programmes in Malaysia and the broader area. Hardy pressured that capability constructing should be inclusive, fast, and strategic.
What comes subsequent
Trying three years forward, Hardy believes many leaders will underestimate the tempo of change. “The primary wave of advantages is already seen in IT operations: agentic AI is automating duties like information classification, storage optimisation, predictive upkeep, and cybersecurity response, liberating groups to concentrate on higher-level strategic work,” he stated.
However the bigger shock could also be on the financial and enterprise mannequin degree. IDC initiatives AI and generative AI may add round US$120 billion to the GDP of the ASEAN-6 by 2027. Hardy sees the implications as broader and sooner than many count on. “The suggests the impression shall be a lot sooner and extra materials than many leaders at present anticipate,” he stated.
In Indonesia, greater than 57% of job roles are anticipated to be augmented or disrupted by AI, a reminder that transformation won’t be restricted to IT. It should lower in how companies are structured, how they handle threat, and the way they create worth.
Balancing autonomy with oversight
The Capgemini findings and Hardy’s insights converge on the identical theme: agentic AI holds enormous promise, however its which means in apply relies on balancing autonomy with belief and human oversight.
The know-how might assist enterprises decrease prices, enhance reliability, and unlock new income streams. However and not using a concentrate on governance, reskilling, and infrastructure readiness, adoption dangers stalling.
For Southeast Asia, the query isn’t whether or not agentic AI will take maintain, however how shortly – and whether or not enterprises can steadiness autonomy with accountability as machines start to tackle extra accountability for enterprise selections.
(Picture by Igor Omilaev)
See additionally: Past acceleration: the rise of agentic AI
Wish to be taught extra about AI and large information from trade leaders? Try AI & Big Data Expo going down in Amsterdam, California, and London. The great occasion is a part of TechEx and is co-located with different main know-how occasions, click on here for extra info.
AI Information is powered by TechForge Media. Discover different upcoming enterprise know-how occasions and webinars here.
