Orbit - An Hg software leadership podcast

Franz Faerber spent 26 years at SAP building Hana before doing something unusual - walking away to build a competitor. Four years in stealth gave Everest Systems the rare luxury of rethinking ERP from first principles, and what emerged is a philosophy that challenges the entire business software model. Faerber's core thesis: application features are no longer defensible because AI can replicate them too easily. It is deep domain knowledge embedded in platforms that enable radical personalization. His "live sandboxing" technology allows users to experiment with production data in isolated environments - something he was never able to do in 26 years at SAP - and represents a fundamental shift from selling configured software to enabling customers to generate their own. The vision is provocative: 80% of software becomes commoditised infrastructure, while 20% becomes bespoke customization, and markets grow as companies build differentiated software instead of accepting generic solutions.

The conversation reveals hard-won wisdom about fighting bureaucracy, the viscosity of enterprise software markets, and why Germany remains his deliberate choice despite Silicon Valley's pull. His analysis of why technology companies struggle to become business software providers offers a refreshing counter to theories around OpenAI or Microsoft absorbing vertical markets. Most personally, his career reflections capture the tension experienced leaders feel between corporate safety and the life-enriching risk of building something new. At an age when contemporaries were settling into senior roles, Faerber chose to start over, arguing AI makes this the first time in ERP history when newcomers can genuinely out-innovate giants in years, not decades.

Daniël Rood, director of AI go-to-market at Google Cloud, offers a masterclass in navigating transformation from inside a tech giant. Building Google's first European AI team when gen AI was still "a PhD science project," Rood reveals the hidden resistance mechanisms that plague even innovative organizations. His concept of the "corporate immune system"—the culture that protects success for all the right reasons but resists dramatic shifts—explains why customer success stories, not internal advocacy, are what actually move leadership. His hiring philosophy centered on "intellectual humility" and "teams of translators" who bridge technology and business offers a blueprint for staffing AI initiatives moving too fast for anyone to be an expert.

The conversation reveals Google's quiet dominance: AI Mode has already reached 1.5 billion monthly active users, nearly double OpenAI's reach. But Rood's most provocative insight addresses vertical SaaS: he predicts the shift to "expertise as a service," where repetitive professional work gets commoditized and deep human judgment becomes "an API on top of the platform." His framework for the new reality is stark—culture trumps strategy, three-year horizons are irrelevant, the world is "tokenizing," and sales cycles collapse through proof-based processes. By 2030, he argues, AI will have moved so deeply into the background that having a Chief AI Officer will seem as obsolete as having a Chief Mobile Officer today.

From oil rigs to Amazon's machine learning division and now Cambridge's DeepMind chair, Professor Neil Lawrence brings a refreshingly grounded perspective to AI. Lawrence introduces his "atomic human" concept; arguing it's not our capabilities but our vulnerabilities and limitations that make us irreplaceable. Drawing on experiences watching his coding assistant try to claim authorship and building systems at Amazon, he illuminates why accountability requires skin in the game and why machines can never truly stand behind decisions the way humans must. His mechanical engineering background shines through in vivid analogies that make complex ideas tangible and even delightful.

The conversation builds to a genuinely uplifting conclusion about the next generation. Lawrence dismisses the disempowering AGI narrative sold by tech incumbents protecting their turf, arguing instead that today's young people see the world as it is and are excited to shape it. His insistence that "people aren't stupid"—from public dialogues to business customers consistently asking for improvements in healthcare and education—makes the case for staying connected to customers and trusting the next generation to steer technology toward what we care about most. It's a perfect note to end the year on: pragmatic, human-centred, and genuinely hopeful.

Product Mind Principal Oji Udezue - veteran of Microsoft, Twitter, Atlassian, Calendly, and Typeform - offers an energising vision of product leadership in the AI era. Drawing from 25 years building products, Udezue reveals that it isn't something to fear but an opportunity for those willing to embrace what he calls "self-erasure": shedding old mental models to approach new technology with fresh eyes.

His journey from Microsoft's competitive trenches to discovering that "permanent influence has nothing to do with convincing people I was smart" unlocks why curiosity and vulnerability now matter more than ever. Udezue introduces the "three speed problem": his framework showing that while engineering capacity approaches infinity over the next decade, the real leverage comes from accelerating product divination and go-to-market to their theoretical limits. With characteristic wisdom, he advocates for "healthy paranoia" about change while championing extreme experimentation, prototyping over PRDs, and designing with customers as core team members. His closing mantra: "I happen to the world; the world doesn't happen to me" captures the determined optimism required to build jewel-like software in an era of unlimited possibility.

In this illuminating conversation from the frontier of AI-native startups, William Fairbairn, founder of Zalos.ai and part of Y Combinator's class of 2025, reveals what it takes to build at Silicon Valley velocity. Just five weeks after launch, Zalos is already deployed with major enterprise customers, automating finance workflows through computer-use agents that execute tasks like humans—extracting contract terms, initiating billing, and reconciling cash.

Fairbairn's insights into Y Combinator's evolved wisdom for the AI era, particularly "forward-deployed engineering", demonstrate how the startup playbook has been fundamentally rewritten. With funding bars rising to $2 million ARR in 12 months just to reach Series A, and "supernovas" scaling from $1 million to $100 million in 18 months, this postcard from the edge of software leadership captures the intensity and opportunity of building in an era where the rules change weekly.