Innovation & Tech

86% of Singapore fintechs deployed AI; only 11% run agents live

86% of Singapore fintechs have deployed AI versus 65% of local banks and 62% globally, yet only 11% run agentic AI in live environments and 9% have implemented IMDA and MAS controls.

By Amara Osei

5 min read

Singapore fintechs lead AI adoption, but struggle to turn usage into business value – survey
Singapore fintechs lead AI adoption, but struggle to turn usage into business value – surveyAI-generated

What's News

  • 86% of Singapore fintechs have deployed AI, against 65% of local financial institutions and 62% of financial institutions and fintechs globally.
  • Only 11% of Singapore fintechs have deployed agentic AI in live environments, though 42% say they are ready to implement it.
  • AI maturity scores: adoption breadth 80, infrastructure and data readiness 67, integration depth 54, autonomy 42, workforce adoption 39.
  • 94% expect AI budgets to increase next year; 41% cite operating cost reduction as the main benefit, only 4% new revenue.
  • 81% engage with IMDA's Model AI Governance Framework and 86% with MAS Project MindForge, but only 9% have implemented controls under each.

By Amara Osei

Singapore's fintech sector has deployed AI at a rate few industries match: 86 percent of companies have already put the technology into production, according to a new report by Kapronasia, produced with global financial technology company Airwallex and the Singapore Fintech Association (SFA).

The same study puts adoption by local financial institutions at 65 percent, and by financial institutions and fintechs globally at 62 percent. Singapore's lead reflects digital-first business models, streamlined infrastructure and organizational agility, the report said.

Breadth of adoption isn't the problem. Depth is.

Kapronasia scored Singapore fintechs across five dimensions. They earned 80 on AI adoption breadth and 67 on infrastructure and data readiness, then fell to 54 on AI integration depth, 42 on AI autonomy and 39 on workforce adoption. "Widespread adoption does not necessarily translate into organizational maturity," the report said.

Nearly six in ten respondents said AI was used by half or fewer of their employees, a gap between tool availability and actual staff usage.

Where the AI actually sits

Usage clusters in the back office. Technology and engineering recorded 99 percent AI adoption, with operations and back-office close behind at 98 percent. High-risk functions lag badly: 59 percent of respondents report no AI deployment in financing, lending and underwriting. The integration gap is wider still — 94 percent in technology and engineering against 29 percent in financing, lending and underwriting. Companies appear cautious about embedding AI where decisions carry significant financial or customer consequences, the report said.

Agentic AI, the class of systems that carry out multi-step tasks and make decisions with greater autonomy, remains rare. Even in the two functions with the highest adoption, deployment stood at only 28 percent in operations and back-office and 26 percent in technology and engineering. Most fintechs remain at the chatbot or copilot stage, where AI assists employees rather than executing entire processes independently.

Readiness, though, is largely solved. 87 percent of respondents consider their infrastructure ready for AI, and 85 percent say their data is sufficiently prepared, against only 23 percent of organizations globally reporting infrastructure ready to scale AI. What's left, the report said, is operational rather than technological: regulatory scrutiny, sensitive financial data and the potential consequences of AI-driven decisions require safeguards and human oversight.

The money question

Budgets are set to grow. Nearly all respondents, 94 percent, expect their AI budgets to increase next year, reflecting opportunities to expand applications and deepen integration across business functions.

Returns are harder to pin down. 61 percent of respondents expected returns on AI investment of 26 to 50 percent, but only 42 percent achieved that range. Meanwhile 45 percent reported actual returns of 1 to 25 percent, against 21 percent who had expected to land there. Conventional return-on-investment measurement may not capture productivity, decision-making and process gains that don't translate cleanly into financial gains, the report said.

Cost cutting dominates the value story. 41 percent of fintechs identified operating cost reduction as their primary AI benefit, while just 4 percent cited generating new revenue. Cost savings and efficiency improvements are easier to attribute directly to AI deployments, whereas revenue growth can be driven by other business factors. Leading fintechs are moving toward function-specific performance measurement to assess AI's contribution, the report said.

Why the agentic gap stays wide

Only 11 percent of Singapore fintechs have deployed agentic AI in live environments, though 42 percent say they are ready to implement it. The report attributed the gap partly to concerns about allowing autonomous systems to operate in sensitive or higher-risk financial functions.

Human checkpoints remain in the plan. 74 percent of fintechs identify human checkpoints as a requirement, and interviews with senior industry leaders indicated human supervision would remain a permanent feature of financial services, particularly where AI agents touch critical financial decisions.

That view isn't universal. Younger start-ups, particularly those with 50 employees or fewer, tend to hold a more AI-native mindset and view full autonomy as a potential longer-term outcome. They recognize the importance of accountability without treating human involvement as the only route to responsible AI, instead seeing trust, appropriate safeguards and governance frameworks as the determinant of how much autonomy AI systems can eventually be granted. The findings point to an evolution in oversight, from human involvement at every stage toward robust controls as the technology matures.

Four blockers to scaling

Singapore fintechs face four main challenges in scaling agentic AI, according to the report: internal data governance, cost, talent and legacy infrastructure. Data governance ranks first, cited by 63 percent of fintech experts. As agents gain access to data and execute tasks across multiple processes, companies need clear controls over what information agents can access, what actions they can perform and when human intervention is required.

Formal frameworks have wide reach but thin implementation. 81 percent of firms are engaging with the Infocomm Media Development Authority's (IMDA) Model AI Governance Framework and 86 percent with the Monetary Authority of Singapore's (MAS) Project MindForge, yet only 9 percent have actively implemented controls under each initiative.

Cost bites as usage scales and token consumption rises; some companies are exploring ways to match tasks with different AI models according to complexity to manage expenses. The talent challenge lies in developing employees with both business knowledge and AI skills rather than simply hiring technical specialists. Legacy technology and fragmented data systems limit agents' ability to access information and execute complex workflows.

Scaling agentic AI will therefore require more than technological advances, the report said, with stronger governance, workforce capabilities, infrastructure and cost management needed to support wider deployment.

The read-through for Singapore's fintech sector is that its AI advantage now sits in tooling rather than autonomous execution. With 94 percent of firms planning larger AI budgets and just 9 percent having implemented controls under the IMDA and MAS frameworks, spending is set to outpace the governance scaffolding around it.

Source: technode.global

Share this article:

More from Amara Osei

Amara Osei

Show full bio

Senior reporter covering companies, deals and strategy in entrepreneurship, startups and venture funding at Business Bearings.

2,248 articles

Related articles

« Previous articleNext article »