A chat with Akin Adegoke and Ayokunle Ilesanmi about trust in the age of Autonomous Finance

In the quiet hum of a modern server room, a credit algorithm can analyse a borrower’s history, assess risk, and disburse capital in three milliseconds, faster than a human loan officer can blink. We are witnessing the dawn of autonomous finance, a world where money moves by code, and decision-making is increasingly delegated to machines.

But as the Nigerian banking sector races toward this frictionless future, a haunting question remains: when the system is on autopilot, who is actually holding the steering wheel?

The shift from manual banking to autonomous systems is not merely a technical upgrade but a philosophical crisis. It challenges the definition of control, the mechanics of trust, and the very nature of accountability.

To navigate this invisible revolution, we turned to two distinct voices at the forefront of the industry: Akinlabi G. Adegoke, Chief Digital Officer of LOTUS Bank, and Ayokunle Ilesanmi, MD/CEO at Berkshire Finance Company LTD. Together, their insights paint a picture of a future that is undeniably faster but one that remains dangerously fragile without the right human architecture.

The paradox of control

For decades, control in banking meant physical vaults and dual-signature cheques. Today, control is spectral, buried in lines of code and API calls. There is a temptation in the tech world to view AI as a set-it-and-forget-it utility, but Mr Adegoke argues that this view is a fundamental misunderstanding of governance.

“At LOTUS Bank, control is multi-layered. It starts with technical oversight, ensuring AI models perform as intended, but extends into governance culture, risk awareness, and regulatory alignment,” he noted. “True control is not just monitoring systems; it’s about embedding accountability and ethical decision-making at every level,” Adegoke explains.

As banking systems become increasingly automated, the risk isn’t just that the machine will break, but that it will work exactly as designed while drifting away from institutional values. Adegoke describes a multi-layered approach at LOTUS Bank where technical oversight is merely the baseline. 

“Humans remain central,” Adegoke insists. “We view our teams as both supervisors and exception managers: they continuously monitor AI outputs, step in when anomalies arise, and guide systemic improvements. AI amplifies human capability; it does not eliminate human judgement.”

This need for human judgement collides head-on with the “Black Box” problem, the tendency of advanced AI models to make decisions through complex pathways that even their creators cannot fully explain. 

Also read: Why Nigeria’s 15.10% inflation number feels like a lie; yet the math says otherwise

Adegoke offers a compelling intersection between ancient financial principles and futuristic tech: “Efficiency cannot come at the cost of trust. We prioritise explainable AI, using models that are interpretable and auditable. In Islamic finance, this aligns with our principles: decisions must be traceable, justifiable, and ethical. Transparency is embedded in both the technology and the culture.”

Ayokunle Ilesanmi agrees, noting, “I think trust depends on context. Banks carry reputational and regulatory weight, which reassures people. Algorithms are trusted when they consistently deliver accurate, fair, and efficient outcomes. Ultimately, people will trust technology more if it’s transparent, explainable, and backed by accountable institutions.”

The consensus is clear: in a market like Nigeria, where scepticism of financial institutions is historical, users will not blindly trust an algorithm. They will trust the institution that can explain what the algorithm did.

Autonomous finance could amplify errors, but

While Adegoke focuses on the ethics of the single decision, Ilesanmi worries about the systemic ripple. Autonomous finance relies on a web of interconnected APIs, cloud infrastructure, and third-party providers.

When autonomy scales, risk does not disappear; it mutates.

“As financial autonomy scales, the main risks are concentration and opacity,” Ilesanmi warns. “Systems may make decisions faster than humans can monitor, and errors or biases embedded in algorithms can propagate quickly. There’s also systemic interconnectedness; issues in one platform can ripple across payments, lending, and investment networks. Strong governance, testing, and transparency are critical.”

This is the nightmare scenario: a flash crash of logic where a flaw in a credit scoring model or a payment gateway doesn’t just affect one customer but cascades through the entire ecosystem before a human can hit the kill switch.

Ilesanmi points out that “Accountability must be traceable, auditable, and clearly assigned. Every participant in the ecosystem – developers, operators, and institutions – needs a defined responsibility. Our current regulatory framework in Nigeria is still largely designed for centralised actors, so updating standards to include distributed, real-time oversight is essential.”

Adegoke agrees, adding, “Ultimately, accountability rests with the institution. AI tools augment decisions but do not replace human responsibility. Our governance ensures clear ownership, and teams are responsible for model design, monitoring, and corrective action when issues arise.”

So, how do we police a system that moves faster than the law?

Ilesanmi proposes a solution that sounds like science fiction but is rapidly becoming a financial reality: cryptoeconomic incentives.

If we cannot monitor every microtransaction manually, we must program the agents to police themselves. “Incentive structures must align rewards with long-term market health, not short-term gain,” Ilesanmi argues.

He envisions a future utilising smart contracts and staking mechanisms that automatically penalise bad behaviour. Instead of waiting for a regulator to fine a bad actor, the code itself could slash the “stake” of an autonomous agent that acts against the market’s interest. It is a shift from external regulation to intrinsic, architectural morality.

However, high-level architecture means nothing if it fails the grandmother in a rural village using USSD code. Both leaders are acutely aware that autonomous finance in Nigeria must serve the financially excluded, not just the digitally native.

Adegoke highlights a critical vulnerability: “Autonomous systems can drift.” Without rigorous testing, an AI model could inadvertently redline entire demographics, effectively automating discrimination. “Preventing bias starts at model development: diverse data sets, ethical review, and ongoing validation,” Adegoke says.

Ilesanmi adds, “Technology can reduce bias, but only if the data and design are carefully managed. There’s a risk of encoding bias at scale if we don’t rigorously audit inputs, assumptions, and outcomes. The focus should be on building systems that are fair, transparent, and continuously monitored.”

As we look toward a horizon five or ten years away, the message from both Adegoke and Ilesanmi is uniform. The banks that will survive the era of autonomous finance are not necessarily the ones with the fastest processors, but the ones with the strongest governance.

“Banks must invest in robust frameworks that combine risk governance, ethical standards, and continuous monitoring,” Adegoke advises. “Most institutions underestimate the importance of culture and accountability structures that evolve with AI complexity, not just technical compliance.”

For regulators, developers, and bank executives, the roadmap is clear: ‘Transparency above all.’ As Ilesanmi concludes, “Systems must be auditable, decisions explainable, and responsibilities clearly defined. When people understand how their money is managed, innovation can thrive without eroding trust.”