Introduction
I walked into my first Regional Scrum Gathering – RSG Hyderabad 2026, expecting the usual Agile refrains: better retrospectives, faster sprints, healthier backlogs. I walked out with something I hadn’t expected at all – a room full of seasoned coaches, CIOs, and delivery leaders quietly asking the same uneasy question in different ways: if AI can now do the work, what exactly are we still here to do?
If there was one sentence that captured the entire two days, it was this: AI handles the output. Humans own the outcome. I heard some version of that idea repeated by nearly every speaker, in nearly every session, from nearly every angle. This post is my attempt to bring that thread together for anyone in our community who couldn’t be in the room.
The Talk That Stayed With Me: The Human Algorithm
Of everything I heard, Anu Smalley’s (CEO of Capala Consulting) session, “The Human Algorithm: Reconnecting Value in the Age of AI” is the one I keep coming back to. Her core argument was disarmingly simple: the heartbeat of innovation has never been the technology – it has always been the people in the room. She named four human capacities that she argued no model can fully replicate: context, ethics, critical thinking, and empathy. Not as soft add-ons to technical work, but as the actual engine of value creation.
What struck me wasn’t the argument itself – most of us intuitively believe this – but how precisely she operationalized it. She wasn’t asking us to resist AI or romanticize the “human touch” as a defensive reflex. She was making a structural claim: technology accelerates the work, but people innovate. Her closing line has stuck with me since: “You are the Human Algorithm. And the world needs you to run.” As someone just beginning to find my footing in the agile community, that felt less like a closing slide and more like a personal charge.
Leadership Gets Rewritten, Not Replaced
Dr. Venkatesh Rajamani, Founder of tryScrum.com, opened the leadership conversation with “The Leader’s New Operating System“. He described every leader as running on an invisible operating system built from their first job, first boss, and first failure – a mix of beliefs, assumptions, identity, and triggers running silently below conscious awareness. His point was that AI doesn’t change your title; it changes your function. The leader who was once the “risk owner” absorbing every failure personally now needs to become a “resilience designer,” building systems that fail safely. The leader who once thrived as the “information broker” now needs to become a “sense maker,” since AI already surfaces the data – the human job is to decide what it means.
Ashutosh Bhatawadekar, a CIO Advisory Consultant and coach, extended this thread by tracing a shift from IQ to EQ to what he called “Gen-AI leadership“, reframing decision-making as a blend of algorithm and intuition. And Dr. Bhadram’s talk, “Do Leaders Laugh?” – probably the most disarming session of the event – asked whether professionals actually perform better when they bring more humor, warmth, and human connection into their work. It was a welcome reminder that leadership transformation doesn’t have to be humorless to be serious.
When the Agent Joins the Sprint
A recurring question was: what happens the moment an AI agent becomes a working member of the team. Vivek Angiras, Co-Founder of Simpliaxis, framed it perfectly: we would never let a new developer commit code on day one without onboarding, context, and boundaries – yet most of our AI agents start working the moment a subscription is activated, with no equivalent onboarding at all. His question to the room – “When an AI agent joins your team, who onboards it?” – was met with a visibly uncomfortable silence, which told its own story.
Gaurav Duggal, CIO at JEPS, pushed this further into more sobering territory with his talk on agentic AI, describing “Multi-agent cascading failures“: agents trusting each other’s output as fact, a single bad decision propagating down an entire chain, and the blast radius growing with every additional connected agent. It was a sober counterpoint to the earlier sprint-planning success stories – speed and reliability, it turns out, are not the same thing.
The Gap Between Pilot and Production
The most quietly alarming data point of the conference came from Ashwini Mathur of Eolas Labs’s session, “The Pilot Worked. The Rollout Didn’t“. She cited MIT research showing that 95% of enterprise GenAI pilots reach no measurable business impact, and a funnel where 60% of organizations evaluate an AI tool, only 20% reach a pilot, and just 5% ever reach production. Her reframe was memorable: the pilot proves the model, but the rollout needs the harness – the evaluation, guardrails, human oversight, and monitoring wrapped around it. Sanjit Bhattacharya’s session, “From Promises to Proof” echoed this with a similar image: AI is the copilot; it belongs in the right seat. Judgment, empathy, and accountability don’t transfer to the tool – they stay human, in the pilot’s seat.
Kamal Raj Sekar, a builder of AI SaaS products, added a sharper edge to this conversation with his talk on AI-induced technical debt. Traditional tech debt, he explained, is something you know you’re taking on – you can see it accumulating, and someone owns it. AI-induced tech debt is different: you often don’t even realize it’s happening, it can accumulate exponentially, and ownership of the code can become unclear because no one fully understands or owns everything the AI generated. That distinction alone is worth carrying back to every engineering standup.
Guardrails Aren’t Optional
Ranjith Rajagopal, an Enterprise Agile Coach, delivered a session on what he called “The cognitive butterfly effect” that was a wake-up call about bias. He cited real, costly examples – the Apple Card credit-limit gender bias investigation, hiring algorithms downgrading women’s resumes for years without detection, and a growing list of multimillion-dollar settlements tied to unaudited AI decisions in lending, hiring, and insurance. His point was structural: teams routinely audit code logic but rarely the data logic or demographic impact behind it. That single gap, he argued, is where the real risk hides – and it belongs squarely inside our Definition of Done, not in a separate compliance conversation months later.
The People Underneath It All
Threaded around these bigger sessions were talks that grounded the AI conversation in everyday career and organizational realities: Kawal Arora, an Engineering Lead at Salesforce, made the case that human readiness has to precede AI readiness, since most transformation failures are people problems wearing technology costumes; Erkan Kadir, a leadership development coach, offered a practical framework for mapping an agile career in a shifting job market; Pramod Sukumaranunni, Founder of Confident Scrum Master, shared advice on standing out as a Scrum Master when hiring itself is being reshaped by AI.
Sudheendra Rayabhagi walked through turning around escalation-heavy delivery in a global capability center, while Natalia Kuzmina and Sandhya Ganapavaram of S&P Global explored driving agility through an AI ecosystem rather than a single tool. And in what might have been the most quietly powerful session of the two days, Ameya Konduru – a teen student leader – shared reflections on building, breaking, and learning through agile practice – a reminder that this community’s future is already in the room.
What I’m Taking Back to My Own Team
Walking out of my first RSG, three things feel non-negotiable for anyone integrating AI into agile work:
- Onboard your AI agents the way you’d onboard a person – with context, boundaries, and a defined role, not blind trust.
- Audit the harness, not just the model – pilots succeed on the strength of the model; rollouts succeed on the strength of the guardrails around it, including bias audits baked into the Definition of Done.
- Protect the Human Algorithm – context, ethics, critical thinking, and empathy aren’t soft skills to preserve out of nostalgia; they are, increasingly, the actual competitive advantage.
So here’s my invitation to the rest of our community: Don’t wait for a tidy answer on where AI fits before you start asking the harder questions out loud – about ownership, bias, onboarding, and what leadership even means when the ground keeps shifting under it.
RSG Hyderabad 2026 didn’t hand any of us a playbook. It handed us better questions, and a room full of people willing to sit with them together. I came looking for practical tips as a first-time attendee. I left looking for the next room full of people asking questions like these – and I’d encourage you to go find yours.