Companies cutting product people are about to learn an expensive lesson about AI.
The bridge between technical capability and business value isn't code. It's the empathy and translation skills you're laying off.
2026 started the way 2025 ended: PM cuts and "open to opportunities" posts flooding LinkedIn.
Three months after these cuts, everything looks fine. Month four is when the wheels come off.
Meta cut 3,600 employees last January. Tidal eliminated its entire PM team. Microsoft shed 15,000 roles. Companies are betting that AI plus engineers can replace PMs.
They're about to discover Claude can write code but can't tell you why users churn.
Building in public forced me to play both roles. These roles aren't interchangeable.
Understanding why requires looking at the math.
Why this looks smart on paper
Product management became an expensive category. In recent years, Senior PMs at tech companies earn $180K-$250K. Multiply that across a team of eight, and you’re looking at $1.6M in annual comp.
Leadership runs the calculation: AI can write specs. Engineers understand technical trade-offs. Design tools prototype interactions. What’s left that justifies $200K per person?
The tasks look automatable. Write requirements. Run sprint planning. Update stakeholders. Collect user feedback. Prioritize the backlog.
ChatGPT handles requirements. Engineers can run their own standups. Product analytics tools surface user data. Jira manages priorities.
On a spreadsheet, cutting PMs while keeping engineers looks like pure efficiency. Maintain technical output while reducing overhead.
Zuckerberg’s memo to Meta employees spelled it out: “We’re moving out low performers faster.” Performance-based cuts sound reasonable until you notice entire PM teams getting eliminated while engineering stays intact.
The pattern repeats across companies. Tidal cut its entire product marketing and product management teams. One engineering leader I follow on LinkedIn posted: “Our PM org went from 12 to 3 in six months. Engineering team unchanged. Executives think AI fills the gap.”
What you’re actually eliminating
Not tasks. Three bridges that can’t be automated or delegated.
When companies cut PMs, they think they’re eliminating coordination overhead. “We don’t need someone to write specs and run meetings.” But PMs weren’t valuable because they scheduled standups.
PMs were valuable because they connected three things that don’t naturally talk to each other: what engineering can build, what customers actually need, and what the business requires to survive. These aren’t coordination tasks. These are translation problems.
AI can automate the documentation. It can’t replicate the judgment that comes from seeing all three perspectives simultaneously and knowing which one should win today.
Bridge one: Technical capability to business value
Engineers optimize for elegant solutions. Business teams optimize for revenue. Someone translates between these languages.
When an engineer proposes rebuilding the authentication system, someone asks: “Will customers pay for better auth? Or do they care about the two features we delayed to do this?”
Building Leafed, AI helped me write better code faster. But AI couldn’t tell me whether barcode scanning or AI-enhanced search deserved the first month of development. That decision required understanding book discovery behavior, competitive positioning, and what would actually get downloads in 2026.
Remove the translation layer, and engineers build technically sound features that don’t move business metrics.
Bridge two: What we can build to what users actually need
Your power users want advanced filtering. Your churning users can’t figure out basic navigation. Which problem do you solve first?
Bangaly Kaba wrote about this for Reforge when he joined Instagram in 2016. The product had 400 million users, but growth had flatlined. Power users loved Instagram. New users couldn’t cross the threshold to become engaged.
Product teams naturally become power users of their own products. This makes them the worst judges of what new users need. AI amplifies this by generating ideas based on existing user feedback. Existing users survived your current complexity.
Who identifies when you’re building for the wrong audience? The PM who can see both what you built and who isn’t using it.
Bridge three: Short-term execution to long-term strategy
Engineers ship code. Executives set vision. PMs connect today’s sprint to next quarter’s positioning.
AI doesn’t understand sequencing. It can’t tell you that Feature A needs to ship before Feature B because A validates the market assumption that makes B worth building.
I’ve seen this pattern on LinkedIn: the PM team is cut, and engineering self-organizes. The first three months look fine. Month four, the backlog runs dry. One engineering manager posted: “Velocity hasn’t dropped. Decision paralysis has quadrupled.”
Building Leafed: Ship barcode scanning first or AI search? Both are technically feasible. Both wanted. Only one could ship this month. That decision wasn’t technical -it was strategic. Which validates our unique value prop? Which builds momentum?
AI doesn’t understand momentum.
Position yourself as AI-augmented, not AI-replaceable
If you’re a PM watching this unfold, your positioning matters more now than your resume.
Generic PM equals replaceable PM. “I write specs and run standups” makes you vulnerable. Those tasks are automatable.
Build your value stack with three layers AI can’t replicate:
Domain expertise that took years to build
You understand SaaS pricing models viscerally. You’ve shipped healthcare products through regulatory approval. You know B2B sales cycles in enterprise software.
That context compounds. AI starts fresh every conversation. You pattern-match across seven years and three industries.
Technical capabilities most PMs don’t have
You can read code and spot architectural decisions that create product constraints. You’ve shipped apps and understand mobile distribution. You know enough SQL to query your own data.
These skills let you work with AI as a multiplier, not as a replacement. You ask better questions because you understand the constraints. You validate AI output because you know what good looks like.
I wouldn’t have shipped Leafed to both app stores without this. AI helped write React Native code. But I needed to understand mobile frameworks, app review guidelines, and distribution channels to know which AI suggestions were actually viable.
Unique perspective your career path created
You went from support to PM and understand customer pain differently. You coded before managing products and spotted technical debt early. You worked in three different industries and see patterns others miss.
Stack these together. You’re not “a product manager.”
You’re “a PM with seven years in fintech who codes and runs a newsletter teaching other PMs, currently building mobile apps to stay technical.”
That’s differentiated. That’s hard to replace with a prompt.
Skills that compound versus skills that commoditize
Commoditizing fast:
Writing PRDs (AI does this now)
Creating mockups (Figma plus AI)
Running standups (async tools replace this)
Stakeholder updates (automation handles this)
Compounding over time:
Reading usage patterns across industries
Recognizing when teams are optimizing local maxima
Knowing which technical decisions create long-term constraints
Understanding when to ignore customer requests
Sequencing bets when everything seems important
Companies cutting PMs bet that the first list matters more than the second. They’re discovering the first list was never the job.
The job was always making better decisions faster than the competition. AI accelerates information gathering. It doesn’t replace judgment built through pattern recognition across hundreds of products and thousands of decisions.
Stop optimizing for task completion. Start demonstrating decision quality.
Final thoughts
Build something that proves you can identify the right problem, sequence the solution, and ship value that users actually want. Newsletter, side project, consulting, open source contribution.
Doesn’t matter which vehicle. What matters: evidence that you can connect technical capability to business outcomes.
Companies that flood LinkedIn with displaced PMs will learn this lesson the hard way over the next 12-18 months. You don’t have to.
Until next week,
P.S. Want to connect? Send me a message on LinkedIn, Bluesky, Threads, or Instagram.
PS: Ready to document before you schedule? I use ClickUp to keep decisions out of my calendar. Grab a free license here and see if it works for your workflow.
This is spot on. AI can handle technical tasks, but it can't replace the human understanding of what users actually need. Great party of people skills!
the month four timeline is spot on. seen this happen before where evrything seems fine until decision paralysis hits