The startups turning Southeast Asia’s AI hype into working agents

The rise of agentic AI – systems that can plan, reason, and act on behalf of users – has emerged as one of the defining stories in tech this year.

Globally, the interest is real: In a 2025 McKinsey survey of nearly 2,000 companies, 62% said they have at least experimented with AI agents. However, those who have deployed agents at scale accounted for less than 10% in any given business function, such as product development, marketing and sales, and HR.

Research and advisory firm Gartner projects that by the end of 2026, 40% of enterprise applications will be integrated with task-specific AI agents, up from less than 5% today.

The cautious pace of enterprise adoption isn’t a surprise to those familiar with the tech’s inner workings. Deploying AI agents in real business environments is more complex and risky than running a pilot.

“Organizations have not shared issues with their agents’ deployment,” Wan Sie Lee, cluster director for AI governance and safety at Singapore’s Infocomm Media Development Authority (IMDA), tells Tech in Asia. “In fact, most of the feedback I got is that people are very careful because they’re concerned that things will go wrong.”

In January, the statutory board published a voluntary framework for agentic AI governance, a guideline for organizations to deploy AI agents responsibly.

The framework, Lee adds, also gives startups a “common language” when pitching to enterprise clients – a way to signal that their agents are “built properly.”

Breaking agents down by category

So what kind of agents are being built in Southeast Asia today?

Tech in Asia’s data shows that 35 startups in the region have raised at least US$1 million building agentic AI products. Most are headquartered in Singapore and targeting enterprises, as B2B lets startups reach enterprise scale through their clients’ existing customer bases.

The largest category by number of startups is functional agents, counting 11 companies building AI agents for specific business workflows, such as marketing, HR, and document processing.

Industry-specific agents – which go deeper into particular sectors like healthcare, construction, and financial compliance – account for 11 companies. Customer-facing agents, where AI directly handles interactions with end users, cover nine startups.

Personal assistants account for four companies, but this category contains the list’s most prominent name: Manus, which Meta acquired in a 2025 deal that valued the AI startup between US$2 billion and US$3 billion. Businesses in this category primarily target individual customers.

The deterministic turn