Can AI really replace software developers - or is it actually exposing how much bad software we’ve tolerated for years? In this episode of The AI Shift, Lee Mallon and James Studdart dig into the real opportunity behind AI: not just building customer-facing products, but creating internal tools, agents and workflows that save time, cut waste, and make teams more effective. From LinkedIn automation and scoping tools to Copilot agents, Power Automate, Buffer, Auth0 and Claude, the conversation explores how AI is changing software development, tech careers, and the economics of SaaS. The hosts also debate whether “manual developers” will be squeezed out, or whether the future belongs to people who can orchestrate AI agents, understand business problems, and turn internal pain points into reusable IP. If you’re interested in AI adoption in business, developer productivity, internal tooling, or what the next phase of tech careers might look like, this one is packed with practical insight and plenty of honest discussion. Chapters: 0:00 Intro 1:03 The AI gold rush and internal tooling 2:34 Quick wins: building tools for yourself and your team 4:03 The manual developer mindset 6:17 Job safety, fear, and AI demos 8:05 Internal IP, tooling, and culture change 10:05 Results-based work versus hours logged 12:06 AI agents, orchestration, and the future of roles 15:17 Salaries, day rates, and the changing market 17:01 Low-hanging fruit for technical teams 19:19 Vibe coding and reviewing AI-generated code 21:33 Pricing, outcomes, and experience 23:40 Internal software optimisation with agents 26:05 Risk agents, Copilot, and training examples 28:53 Learning to ask the right questions 31:09 Rewiring how we solve problems 33:22 Bias, technical mindset, and solutionising 36:23 Why internal tools should be built in-house 39:08 SaaS pricing, feature bloat, and Buffer 41:44 Auth0, authentication, and pricing models 47:30 Building products from internal tools Notable quotes: “Why keep doing those sort of things and use your token limits when you could have your token limits used once and build something to repeat it?” “Job safety is a fallacy. It doesn’t exist.” “Those people that can just write the code, they’re not going to earn the salaries that they’re earning now.” “The people that can orchestrate those AI agents and things, that’s where the money’s going to be.” “Why do I make it easier for them to dismiss me and bring someone else in that could just use software to run my business?” Resources mentioned: Claude LinkedIn Scope Lock T4 templates Microsoft Copilot Power Automate MIT risk database Buffer Auth0 Figma TikTok Excel COBOL Pascal Python Subscribe to The AI Shift for more real-world AI discussions with Lee Mallon and James Studdart, and follow us for future episodes on AI, software development, and the changing world of tech.
