More Software Engineering Radio episodes

Marian Kamenistak on Product Management for Software Engineers thumbnail

Marian Kamenistak on Product Management for Software Engineers

Published 12 Nov 2026

Duration: 01:02:24

As AI automates coding, engineers must master product management, focusing on customer value, rigorous validation, and business outcomes to remain valuable.

Episode Description

Marian Kamenistak, founder of the Engineering Leaders Community and former VP of Engineering at Mews, joins host Priyanka Raghavan to discuss product...

Overview

The conversation explores how product management is becoming an essential skill for software engineers as AI accelerates code generation. The guest argues that the distinction between solving technical problems and solving customer problems is largely artificial - technical improvements must ultimately deliver value to a customer or the company itself. With AI making development "the boring part," the critical work shifts to what happens before development (understanding customer motivation, creating right specifications) and after development (activation, adoption, and measuring outcomes).

The discussion covers practical case studies, including a database scalability challenge at Databricks where accepting near real-time performance instead of true real-time eliminated the need for complex replication architectures, and an insurance company that built elaborate dashboards only to find customers exclusively used a "download Excel" button. The guest identifies code review speed as the emerging bottleneck, proposes metrics like delivery cycle time, roadmap contribution percentage, and production defects (rather than total bug count), and introduces the concept of "product builders" - individuals who cover the full product lifecycle from customer discovery through marketing and revenue measurement.

The conversation also addresses how product specifications change when the customer is an AI agent rather than a human, requiring stricter availability guarantees and validation workflows because probability-based systems cannot be trusted to follow patterns without explicit constraints.

What If

  • What if you treated your next feature spec as a customer-contract negotiation before writing a single line of code?

    • Move: Sit with a customer (or proxy) and transcribe the conversation into an AI-assisted specification covering both functional and non-functional requirements (throughput, latency, security, stability, high availability), as the guest describes for spec-driven API development.
    • Why Now?: The guest notes that "even a product manager who is not deeply technical can build an exquisite API these days in the year 2026" using AI tooling for spec-driven development.
    • Expected Upside: You may discover, as in the Databricks case, that near real-time suffices for the customer's use case, potentially avoiding over-engineering and saving significant development time. This is a possibility, not a guarantee.
  • What if you measured your product by production incidents rather than internal bug count?

    • Move: Set up observability and telemetry to track production defects, customer complaints, and support ticket spikes as your primary quality signal, and establish a deployment freeze threshold (e.g., if incidents rise from 5% to 30% of tickets, halt new features and fix foundations first).
    • Why Now?: The guest states that "the number of defects or incidents in production... has a direct impact towards the customer as opposed to measuring the full back of bugs", and notes AI can now auto-resolve many incident tickets overnight.
    • Expected Upside: You gain a concrete, customer-facing metric to justify pausing feature work, which the guest says is essential before shipping new functionality. This is a proposed approach, not a universal rule.
  • What if you positioned yourself as a "product builder" rather than a single-role developer?

    • Move: Package your skills into three tiers - core technical skill, product/communication/data skills, and AI workflow automation - and present this portfolio to clients or employers, including how you will handle activation, adoption, and revenue measurement.
    • Why Now?: The guest predicts that by 2027 and 2028, forward-deployed engineers and product builders who understand marketing and sales will be "the most demanded on the market", and that "software development becomes the boring part".
    • Expected Upside: You differentiate from engineers who only press "yes" to AI suggestions, positioning yourself as someone who covers the full lifecycle from customer discovery to profit generation. This is a proposed positioning strategy, not a guaranteed outcome.

Takeaway

  • Ask customers "why" five times before designing a solution; the guest recommends this to reach the root cause of what they actually need, which may eliminate complex technical requirements entirely.
  • Before shipping a new feature, check whether production incidents have spiked (e.g., from 5% to 30% of tickets); if so, freeze deployment and fix quality minimums first.
  • Measure engineering success by delivery cycle time (months vs. weeks vs. days), roadmap contribution percentage, and production defects - not total bug count.
  • When building APIs or integrations, use AI-assisted spec-driven development to transcribe customer conversations into specifications covering functional and non-functional requirements (throughput, latency, security, stability, high availability).
  • Present yourself as a portfolio of skills (core technical, product/communication, AI workflow automation) rather than a single role, and include how you will handle activation, adoption, and revenue measurement.

Recent Episodes of Software Engineering Radio

29 Oct 2026 Danilo Krummrich on Rust in the Linux Kernel

"Nova GPU driver project integrates Rust into the Linux kernel to replace the aging Novo driver, leveraging Rust's memory safety for better reliability and performance while addressing firmware and architectural challenges through incremental, collaborative updates."

7 Oct 2026 Sahil Walia on Schema Changes and Data Migrations

Recommended: Schema as source of truth

"Schemas define data structure, apply across systems, and require careful evolution to avoid breaking changes, with best practices including additive updates, blue-green deployments, and stable identifiers."

30 Sept 2026 Raju Dandigam on Building Production AI Agents

"Explores production AI agents, their autonomy, safety, and engineering best practices, comparing workflows, tool design, orchestration, and memory systems while addressing challenges like cost, latency, and reliability."

16 Sept 2026 Milan Milanovic on the Laws of Software Engineering

"Explores key software engineering principles (like Conway's Law, Brooks' Law) and their impact on systems, teams, and decision-making, emphasizing context-dependent trade-offs, AI's role, and practical applications like measuring technical debt."

9 Sept 2026 Owen McGirr on Software Accessibility

"Accessibility in software development must be prioritized from the start, integrating inclusive design practices like multiple input methods, proper labeling, and user testing to benefit all users, not just those with disabilities."

More Software Engineering Radio episodes