Best of LinkedIn: Digital Products & Services CW 44/ 45 2026

AI in the Product Manager's Workflow

• Figma runs a 31 skill PM OS built on Claude Code from one shared git repo, keeping 60 to 70 product managers in sync
• Polsia's agent setup logs customer issues during conversations, ranks the day's tickets and passes the top items to engineering agents that build the fix
• As AI agents speed up coding, the bottleneck shifts from writing code to requirements, decisions, verification and judgment
• AI absorbs PM administrative work, while accountability for decisions stays with the human
• Product teams ship faster with AI, yet customers gain little until product practices adapt to the new pace
• In design workflows, lead and designer pairs give way to expander and refiner pairs
• Strategy decks are turning into interactive, evidence-backed landing pages built with AI

‍

AI Product Strategy and ROI

• In ERP, AI value comes from understanding the customer's industry context, not from shipping more agents
• AI features earn adoption by naming outcomes for a specific customer segment, not by claiming a generic assistant
• An AI-native bank shows the pattern: valuable AI features solve a real user job end to end
• A 56% increase in tasks completed is a productivity metric, not AI ROI. The return has to show up as revenue growth
• Asana's CPO describes delivery cycles collapsing from years to weeks to hours, which makes judgment and ROI alignment the defensible moats. Studies cited put roughly 95% of genAI pilots below measurable business impact, with CFOs allowing 12 to 18 months of experimentation
• Google's Gemini product head frames the PM role as shifting from writing requirements to building product with heft, and says a first eval can be five prompts
• AI shopping discovery still rests on product quality, accurate attributes, reviews and availability

‍

Product Operating Models and Organization

• Most companies run an undesigned product operating model that accumulated over time rather than being chosen
• Product operating model adoption fails when organizational conditions and service layers are left unaddressed
• AXA's One GTO Tech Factory brings Products, Service Operations and Markets into one industrialized model under three priorities: refocus, industrialize, accelerate
• Product Operations gains weight as AI speeds delivery, supplying visibility, alignment and outcome focus
• Transformation Theater layers new tools onto old systems without changing how decisions are made
• Product carries accountability for outcomes while other functions control most of the decisions that shape them
• Product commonality savings erode through engineering changes and deal by deal customization, and sales KPIs keep that cost invisible

‍

Product Discovery and Customer Proximity

• The shift under discussion is from tracking project outputs to owning business outcomes
• Product managers should own the full loop from customer conversation to delivery
• Product Owners need direct customer access, not requirements handed down without customer context
• AI now extracts repeated problems, feature requests and frustrations from customer calls for product teams
• A ProductFest 2026 panel argued that discovery was already broken and AI worsens it, as teams skip to ideation and iterate toward a less bad first idea
• A Business Decision Brief links discovery findings to planning decisions and tradeoffs
• AI returns product managers' time to customers, empathy and judgment

‍

Roles, Careers and Hiring

• Levels.fyi data puts PM to engineer ratios at 1:30 for Apple and 1:3.6 for Walmart, but the figures come from US submissions over 24 months, not verified headcounts
• The same Levels.fyi data shows PM to engineer ratios varying widely across companies
• In a review of 33 product owner job postings, 16 of 19 internal product roles expected the owner to decide what to build and why, while 9 of 13 roles for sold products centered on backlog management
• Six types of product managers serve different missions, so hiring should reflect those differences
• AI product management extends the traditional role with responsibility for AI quality, evaluation and tradeoffs
• Skill assessments completed by more than 21,000 PMs identify eight critical knowledge areas across roles from Product Owner to CPO
• Product careers move through three phases: growing seniority, the product leader canyon and the step from manager to executive

‍

SaaS Economics and Market Moves

• Bending Spoons acquiring Miro raises doubts about the future of the whiteboard category
• The product lead of Claude Code is joining OpenAI's Core Products group as Anthropic nears a Nasdaq listing, and OpenAI agreed to buy Pierre, whose Git tools already run inside Codex
• ThoughtSpot was named a Leader in the 2026 Nucleus Research Embedded Analytics Value Matrix and ranked highest on usability
• ProdLucent launched as a SaaS tool connecting corporate strategy, initiatives and software development roadmaps
• SaaS products that AI can rebuild in a weekend face collapsing subscription models, including those priced around $29 a month
• The counterargument: if 1,000 customers each replace a SaaS product with AI-built software, each needs one or two developers to maintain it, which can exceed the vendor's own development cost
• Embedded finance in SaaS calls for a product roadmap, not just additional payment providers

‍

Product Craft and Practice

• A traffic mix shift hid working experiments. Winners converted 8% better than the holdout while overall conversion fell 16%, because paid traffic grew from 10% to 35% of page views. At constant mix, conversion rose 7.5%
• The MoSCoW framework prioritizes feature requests, illustrated with a scan to pay launch
• Habits borrowed from strong PMs include simple PRDs, decision logs and close engineering relationships
• Product sense is the ability to find the right user and business solution amid ambiguous information
• Google Maps shows how one experience can serve users in very different contexts
• Outages at payment, hosting and DNS providers show why a tested fallback belongs in the plan from day one
• Building a personal app end to end with AI sharpens product management skills

‍

Thanks to Claire Vo, Ben Cera, Sean G., Nayan Teja, Roman Pichler, Bryan Zmijewski, Dmitrii Krupenin, Jennifer Sherman, Tim Herbig, Dr Bart Jaworski, Vin Vashishta, Grace Gong, Lenny Rachitsky, Seth Hurd, Melissa Perri, George Sheldrake, Hervé Le Hen, Jasleen Kaur, Jen Swanson, Stephanie Leue, Sergei Krasulya, Matthieu Sors, Michael Heiberg, Summer Lawrence, Kapildev Arulmozhi, Beth McHugh, Amy Mitchell, Laura Shores, Sabah Qazi, Kunal Dudhraj Biswakarma, Kent McDonald, Anand Sagar, Muhammad Zain ul Abadin, Paweł Huryn, Aakash Gupta, Christoph Steinlehner, Melissa Rosenthal, Brian Reynolds, Kurt Johnson, Alex Taillandier, Tom Collopy, Jim Fitzgerald, Ravi Mehta, Unwana Brownson, Farah H., Prasad Reddy, Omar Sallam, Zain K. and David Barnett and everyone else who contributed insights to this edition.

Find the full list of posts and voices on LinkedIn: https://www.linkedin.com/pulse/best-linkedin-cw-44-45-digital-products-services-thomas-allgeyer-fk35e/