LC Ideas: Views & Insights
The rise of AI agents and what does it mean for the financial services industry?
Table of Contents
On January 27, markets reacted with surprise to a sudden shift in the AI investment narrative. DeepSeek’s new AI model delivered state-of-the-art performance while achieving major efficiency breakthroughs—contradicting the assumption that AI progress would require exponentially more computing power. With DeepSeek emerging as a key player so swiftly, what does this signal for the AI industry?
"With DeepSeek coming in, AI is becoming cheaper and more scalable," said Shilpi Chowdhary, Group CEO of Lighthouse Canton. "This is going to be a game-changer for businesses looking to adopt AI without prohibitive costs."
The arrival of DeepSeek is dramatically altering the cost dynamics of artificial intelligence, making advanced AI capabilities more accessible and accelerating adoption across industries. Its swift rise highlights its transformative potential. In just days, it climbed to the top of U.S. app store charts as the leading free app, inspired over 700 (and counting) open-source spin-offs, and gained integration into major platforms like Microsoft, AWS, and NVIDIA's AI ecosystems.
BUILDING TASK-SPECIFIC AI AGENTS
The efficiency gains unlocked by models like DeepSeek are fuelling a parallel revolution: the rise of agentic AI. Unlike general-purpose models, these autonomous agents are designed to execute specific tasks—analysing data, automating workflows, or even making decisions—with minimal human intervention. Enterprises are increasingly adopting them to replace rigid, rules-based systems with adaptive, learning-driven tools.
For investment firms like Lighthouse Canton, the shift is particularly transformative.
"AI agents aren’t just chatbots—they’re domain-specific problem solvers," Chowdhary explains. "The cost barrier was once a hurdle, but now we can deploy them at scale."
By the end of 2025, Chowdhary anticipates that Lighthouse will have 10 to 15 AI agents performing specialised tasks.
Among these is a conversational portfolio tool that fetches real-time data for portfolio performance, risk assessment, and corporate actions. Unlike traditional LLMs, which can struggle with specific financial queries, these agents provide dynamic, precise responses.
Another notable development is the KYC agent, designed to transform the often tedious KYC process into an engaging experience. "Imagine a system that asks for your name and then engages in a dialogue, asking you questions about you. By the end of the interaction, we’ve completed KYC in a more user-friendly manner," he shared.
Beyond these, Lighthouse Canton is also working on transaction monitoring agents, operations and marketing automation, and portfolio forecasting tools in collaboration with educational institutions like NUS. Additionally, a customer support automation agent is in development to streamline user interactions by reducing repetitive queries and offering personalised assistance.
Chowdhary also sees AI agents revolutionising risk management in financial markets.
"We're moving towards AI-powered compliance tools that detect anomalies in financial transactions in real time, flagging potential risks instantly," he noted. This level of automation is particularly critical in highly regulated industries, ensuring organisations remain compliant while improving efficiency.
AI’S EXPANSION BEYOND THE US
The AI landscape has long been dominated by U.S.-based developments, but Chowdhary observes a shift.
"Markets like Singapore and Dubai are emerging as innovation hubs," he said. This is particularly relevant in the financial sector, where AI applications are tailored to heterogeneous regulatory environments.
Unlike the US, where a homogeneous system allows for seamless AI adoption, Asia presents a different challenge.
"In the U.S., you have a $27 trillion economy operating under one system. In Asia, every few thousand kilometers brings a change in borders, regulators, currencies, and cultures. This complexity makes AI adoption more intricate but also presents unique opportunities," he noted.
He also highlighted the significance of government-led AI initiatives in Asia, particularly in digital banking and fintech.
"Singapore and the UAE are setting up regulatory sandboxes to test AI-powered financial solutions. These controlled environments allow innovation while ensuring regulatory oversight," he shared.
Industry data supports this shift.
According to a report by IDC, AI spending in Asia-Pacific (excluding Japan) is expected to reach $49.2 billion by 2026, reflecting a compound annual growth rate (CAGR) of 24.5% from 2022. The increased investment highlights the region’s growing focus on AI-driven financial solutions and enterprise automation.
THE RISKS AND OPPORTUNITIES IN AI ADOPTION
Despite the promise AI holds, Chowdhary remains pragmatic about its risks.
"AI is obviously the buzzword everywhere, but real applications will take longer to mature than most expect. Large organisations move cautiously due to concerns over data security and regulatory compliance," he shared.
One of his key concerns is the potential misuse of AI. "The technology brings remarkable productivity gains, but it also enables malicious actors. Deepfakes, voice replication and AI-powered scams are growing risks."
He emphasised the role of cybersecurity in mitigating these threats. "The challenge is that AI is evolving faster than regulatory frameworks. Governments and institutions must catch up."
He also warned of bias in AI decision-making, particularly in lending, insurance, and recruitment. "If AI models are trained on biased data, they will produce biased outcomes. Organizations must implement fairness checks to ensure ethical AI applications."
Looking ahead, Chowdhary identified robotics and pharmaceuticals as two sectors where AI will drive significant innovation.
"Robotics will expand in both industrial and consumer applications, while pharmaceuticals—historically led by large US firms—could see greater innovation from Asia."
With AI adoption set to surge and new cost efficiencies making the technology more accessible, businesses and regulators alike must navigate the challenges and opportunities that lie ahead. "We are at the very early stages of AI’s evolution. The real challenges—and breakthroughs—are yet to come," Chowdhary concluded.
