Finance Upturned Podcast

Episodet

• 
  AI has changed banking forever, analysts need to perform like experienced bankers on day one 03.09.2026 46min

  For decades, junior bankers learned by doing much of the grunt work in their teams. This included tasks such as spreading financial statements, printing and analysing tax returns, calculating ratios and working through large volumes of financial information.The work was often tedious, but it gave junior bankers time to understand how businesses worked and, importantly, where risk could hide in the numbers.AI is now capable of doing much of this work in minutes rather than hours. That creates a difficult question for banks: if junior analysts no longer spend their first few years doing the work through which they traditionally learned credit, how do they develop the judgement needed to become experienced bankers?In this episode of Finance Upturned, I speak to Joshua Summers, a founder with two exits, and co-founder and CEO of EnFi, an agentic AI platform for banks and private lenders.Joshua and his co-founders spoke to around 100 lenders before building EnFi and concluded that credit risk itself was not the problem. The problem was having enough people to do the work required across origination, underwriting and portfolio monitoring. EnFi’s proposition is to help lenders scale without scaling headcount.We discuss how agentic AI is being applied to complex credit, why humans still need to remain involved in the process, and what happens to junior training when AI takes over much of the manual work.We also discuss the regulatory requirements around AI in credit, the risks of “vibe coding”, whether AI will actually reduce headcount, and what young people should be learning as AI changes the workplace.Best quotes“Credit risk is a pretty well understood science. The challenge is actually in people and the capacity to do that work.”“We can’t just have humans all of a sudden be a chief credit officer without having learned anything along the way. That would be disastrous.”“Now a single underwriter is able to do six to eight loans a day, where before they were doing one to two.”Timestamps* 08:34 How Joshua explains agentic AI to experienced banking executives* 19:02 The three things regulators need from AI: auditability, explainability and provenance* 21:24 Why complex commercial credit is a good fit for agentic AI* 24:11 How AI can cut a 16–32 hour annual review process to a couple of hours* 25:11 SBA underwriting cut from hours to 30 minutes* 30:08 The hidden danger of “vibe coding”* 35:27 How AI could change the way junior bankers learn credit* 38:20 What Joshua would advise a young person entering the workforce today* 41:37 Would Joshua go into software engineering today* 43:41 Why AI-generated content could push people back towards more human interactionGuest: Joshua Summers, CEO & Co-founder, EnFiHost: Kayode Odeleye, Finance Upturned Get full access to Finance Upturned at kayodeodeleye.substack.com/subscribe
• 
  Why 95% of AI projects fail: Accenture's David Hardoon on people over technology 21.07.2026 26min

  David Hardoon is today one of the best known names in AI, but it all started from detention, 3 decades before AI became a buzzword. As punishment for being an errant teenager, he was sent to the school library, where he found a book on Prolog which piqued his interest in technology.After an undergraduate degree in computer science with AI and a PhD in machine learning, he worked in data, analytics and AI, before being appointed as the first Chief Data Officer of the Monetary Authority of Singapore. A long stint in charge of data and AI followed at UnionBank of the Philippines and the wider Aboitiz group, and more recently as Global Head of AI Enablement at Standard Chartered. Just last month, he was appointed Managing Director and Head of Advanced AI for Southeast Asia at Accenture.He’s also the author of The AI Failure Handbook. For someone with his profile, you’d expect a textbook or a treatise on AI adoption. Instead, David published a satirical guide fashioned after a 1944 wartime sabotage manual. Which fits how he thinks about AI: start with the people.In this conversation, that thread runs through everything. His answer to why most AI initiatives fail is that leaders treat AI as technology, the CEO turns to the CTO and says make it happen, and nobody thinks about the people who have to live with the result.We talk about:* Why his AI strategy is the business strategy, and what that means in practice* The financial crime deployment that worked because the team designed it with the domain experts, down to what happens when a customer’s account gets flagged* Layoffs and the boomerang: why he thinks firms that cut for efficiency will rehire, and what Klarna’s reversal signals* Keep the people, do more: his case against using AI to do the same work with fewer staff* Gen Z, graduate hiring and the succession problem nobody prices in when they automate junior work* What Singapore’s government is doing about workforce readiness that other jurisdictions could copy* His advice to a 21-year-old: the thinking process matters more than the tools, and read philosophy* Using AI to write when you’re dyslexic, and the question that settles whether it’s cheating: whose idea was it?* Whether the data centre buildout is a bubbleDavid is the author of The AI Failure Handbook: The Ultimate Guide to Screwing Up Your AI Transformation, available on Kindle and in paperback. Get full access to Finance Upturned at kayodeodeleye.substack.com/subscribe