Finding Balance in the Chaos: Processes that Empower TPMs in the Age of AI

In the whirlwind of startup life, the right processes can be a lifesaver. I reflect on incident management, SLO/SLA hygiene, and more, exploring how to avoid anti-patterns while embracing the agility that AI brings to Technical Program Management.

Balancing Chaos And Innovation Daily

It was a Tuesday like any other in our bustling startup, the kind of day where the coffee is strong, the deadlines are tighter, and the chaos is palpable. I sat at my desk, staring at the multitude of screens, each populated with alerts, metrics, and a growing pile of tickets. The moment felt like a microcosm of our journey: a blend of innovation and unpredictability. As a Technical Program Manager (TPM), I often find myself juggling not just tasks but the very processes that keep us on track amidst the storm.

In this world of rapid iteration and agile methodologies, the importance of robust yet flexible processes cannot be overstated. While we often find ourselves leaning into the allure of speed—especially with the advent of AI automation—we must also be vigilant against the creeping vines of bureaucracy and cargo-cult practices that can stifle our innovation.

One of the most critical processes we engage in is incident management. I remember a particularly challenging incident last quarter when our platform went down for a few hours, sending teams scrambling. In the aftermath, we embraced a blameless postmortem approach, which allowed us to dissect the incident without finger-pointing. By focusing on systems and processes rather than individuals, we cultivated an environment where team members felt safe to share insights and learnings. This not only improved our response times for future incidents but also fostered a culture of continuous improvement.

However, good intentions can sometimes lead us astray. I’ve witnessed organizations fall into the trap of bureaucratic processes that slow down the very agility we strive for. For instance, when teams get bogged down in excessive documentation for every incident report, we risk creating a culture of fear rather than learning. It’s essential to strike the right balance; our postmortem rituals should be lightweight yet effective, allowing us to learn without overwhelming our teams.

Speaking of balance, let’s talk about SLO (Service Level Objectives) and SLA (Service Level Agreements) hygiene. These metrics are vital to our operational health, but often, they become a checkbox exercise rather than a meaningful reflection of our service quality. When we define SLOs, we need to ensure they are not just aspirational but truly informed by data and aligned with user expectations. This requires ongoing dialogue between engineering and product teams, and the agility to adapt our objectives based on real-world performance. AI can play a crucial role here, helping us analyze user behavior and service reliability, allowing us to adjust our SLOs in real-time.

As we explore the intersection of governance and speed, I’ve found that adopting release trains and quality gates can be game changers. Establishing a regular cadence for releases—like our bi-weekly release train—ensures that we maintain momentum without sacrificing quality. Yet, I’ve seen teams fall into the cargo-cult trap, adhering to rigid processes without understanding the underlying principles. It’s not just about the process; it’s about what we’re trying to achieve with it. Each quality gate should be a thoughtful checkpoint, not a hurdle that stifles innovation.

Our design and

Transforming Feedback: From Dread To Collaboration

PRD (Product Requirement Document) review rituals are another area where I’ve seen both triumphs and pitfalls. Initially, we approached these reviews with a heavy hand, creating an environment where teams dreaded the feedback sessions. But by shifting our focus to collaborative design thinking, we encouraged open discussions that fostered creativity. The introduction of AI tools for design validation has further streamlined this process, allowing us to gather insights and iterate faster while still honoring the essential feedback loops.

Ultimately, the processes we choose to implement should feel like a guiding framework rather than a straitjacket. As TPMs, we have the opportunity to shape these processes in ways that not only drive efficiency but also empower our teams. Healthy patterns emerge when we embrace a mindset of adaptability and data-informed decision-making. We must remain vigilant against the pull of bureaucracy and the allure of cargo-cult practices, ensuring that our processes serve our goals rather than hinder them.

As I wrap up this reflection, I am reminded of a quote from the world of AI: "The best algorithms are those that learn and adapt over time." This should resonate with us as TPMs. Our processes, too, must be seen as living entities—constantly evolving to meet the needs of our teams and our users. In the chaos of a startup, it’s not about having the perfect process; it’s about having the right processes that allow us to navigate change with resilience and agility.