daily

The Dance of Processes: Finding Harmony in Chaos with TPM and AI

In a world swirling with hype around AI and generative models, a skeptical TPM reflects on the dance of processes—balancing governance and speed, and embracing healthy patterns over anti-patterns. Join this journey through incident management, SLO hygiene, and more.

TPMxAI

24 Jun 2025 — 3 min read

The Dance of Processes: Finding Harmony in Chaos with TPM and AI

Navigating Chaos In Tech Management

As I sat in a dimly lit conference room, the glow of screens illuminating the faces of my colleagues, I couldn’t help but feel we were trapped in a strange rhythm of chaos. The latest AI model had just rolled out, and with it came the inevitable hiccups. My mind drifted, contemplating the underlying processes that define our work as Technical Program Managers (TPMs) in this rapidly evolving landscape. It felt like a dance, both exhilarating and terrifying, where each misstep could lead to a cascade of errors.

Incident management was our first partner in this dance. The promise of a blameless postmortem was like a siren’s call, beckoning us to come together in the aftermath of a failure. Instead of pointing fingers, we gathered around the metaphorical campfire, sharing insights and learning from our mistakes. I remember the first postmortem I led—a failed deployment that sent our service crashing. Instead of laying blame, we dissected the incident, highlighting the gaps in our SLO (Service Level Objective) and SLA (Service Level Agreement) hygiene. It was cathartic and transformative, turning our failures into lessons rather than scars.

Yet, as we embraced this practice, I couldn’t shake the feeling that some of us were slipping into a dangerous groove—cargo cult behaviors that mimicked the rituals without understanding their essence. Too often, I witnessed teams adopting ‘postmortem’ rituals as a checkbox exercise, where the focus shifted from learning to merely documenting. The beauty of a blameless postmortem lies in its potential for real growth, yet here we were, swaying to a tune devoid of meaning.

Our release trains, too, were an intricate choreography. The rhythm of planning, development, and delivery was supposed to create a seamless flow. But in practice, I saw teams bound by excessive governance—rigid quality gates that stifled creativity and speed. I often wondered if we were building barriers instead of bridges. I recalled a particularly painful release cycle where the gatekeepers had grown so cautious that we lost momentum, leaving us feeling more like bureaucrats than innovators. In the world of AI, where speed is the name of the game, we must find a balance between governance and velocity.

As I navigated these challenges, I learned the importance of lightweight, data-informed processes. A friend once compared effective process management to a jazz band; each member knows their part but is free to improvise as needed. SLOs became our sheet music—guiding us but allowing for creativity in execution. In one instance, we adjusted our SLOs based on real user feedback rather than arbitrary metrics, leading to a significant increase in customer satisfaction. It was a beautiful moment, a reminder that data should inform our decisions, not dictate them.

Then there are the design and PRD (Product Requirements Document) review rituals. These can become lengthy and tedious affairs, with stakeholders circling like hawks, eager to impose their perspectives. But I’ve found that keeping these reviews collaborative and focused on outcomes can foster a healthier dialogue. I recall a recent PRD review where we facilitated a workshop instead of a traditional meeting. By breaking into small groups, we encouraged open discussion, leading to richer insights and a more refined product vision. It was a stark contrast to the previous sessions, where we’d often drown in a sea of opinions and end up with a bloated document that pleased everyone yet satisfied no one.

Yet, the dance isn’t all smooth. I’ve seen the anti-patterns emerge, like unwanted guests crashing a party. Bureaucracy can creep in, suffocating our processes under the weight of unnecessary approvals and documentation. Cargo cult mentalities take root, where teams follow procedures without understanding their purpose. It’s a slippery slope, and as TPMs, we must remain vigilant. We need to nurture an adaptive culture—one that values agility and responsiveness over rigidity.

In this era of AI, where hype often overshadows reality, the role of the TPM becomes more critical. We stand at the intersection of technology and strategy, tasked with navigating the complexities of innovation. Our processes are not just bureaucratic hurdles; they are the frameworks that allow us to harness the potential of AI effectively.

As I reflect on our journeys through incident management, release trains, and review rituals, I see a tapestry of experiences woven together by our commitment to improvement. Each misstep, each triumph, contributes to our collective knowledge. In the end, it’s not just about the processes we implement but the culture we cultivate—a culture that embraces learning, values speed, and understands the delicate balance of governance.

And so, as the conference room buzzed with the energy of ideas and discussions, I realized that the dance of processes is ongoing, a rhythm we must continuously refine.