cases

In February 2024, a small-claims tribunal in British Columbia ruled that Air Canada was legally on the hook for a refund policy its chatbot invented. The dollar amount was trivial. The precedent rewrote the risk model for every AI product team shipping a customer-facing assistant. This is a case study about what happens when a hallucination meets a contract — and what the legal record now expects you to have built before launch.

How a couch-share idea evolved into a global category — the leadership decisions that compounded.

In January 2025, a Chinese quant fund's side project erased $589B from Nvidia's market cap in a single trading day. The real story isn't that DeepSeek beat OpenAI — it's that the AI industry's capex narrative had been wrong for 18 months and almost nobody outside DeepSeek noticed.

Copilot is the most successful AI product launched to date. The lessons from its five-year arc are buried under a pile of marketing claims that the underlying research never actually made. A case in distribution, UX wedge, and what productivity research really says.

Klarna's February 2024 AI announcement was a masterclass in PR — two-thirds of chats handled, 700 agents' worth of work, $40M of projected profit, all in the first month. The 2025 walk-back was a masterclass in judgment, and it is the part every AI PM needs to study harder than the launch.

While most B2B SaaS bolted a chatbot onto the corner of its product, Linear threaded AI into the surfaces where decisions already happen — the comment thread, the cycle review, the triage queue. A case in AI feature restraint inside a product whose brand is restraint.

AirAsia's "Now Everyone Can Fly" strategy wasn't just marketing. It was backed by a deliberate IT infrastructure built from 2001 that made low-cost operations viable at scale. A case in technology enabling business strategy.

Amazon's recommendation engine, anticipatory shipping, and dynamic pricing are not add-on features — they are the product. A case in data science as the primary product strategy at scale.

How Anthropic built a consumer product around a foundation model they also own — the PM decisions when the model team and the product team share the same building, and the trust, fallback, and tool-use design that defined Claude's launch.

Apple didn't invent the MP3 player. It read the market correctly, made one decisive integration bet, and turned a crowded hardware category into a decade-long category leadership position.

Blinder was a dating app that tried to match people on personality and hobbies, not photos. Its UX didn't survive first contact with users — but the research process behind it is worth studying closely.

How Box scaled from a consumer file-sharing startup to an enterprise content platform — and what the cultural decisions along the way tell leaders about the relationship between product identity and organizational design.

Circa launched in 2012 with a genuinely novel approach to digital news — atomic, object-oriented stories. It shut down in 2015. The idea was sound. The business model wasn't.

A business school professor bought an online lingerie affiliate site for $30K and turned it into a retail operation. Six weeks after relaunch, business was almost non-existent. A case in e-commerce metrics and growth diagnostics.

How a two-year-old startup wrapped an LLM API into a defensible coding product that challenged GitHub Copilot — and what the build/wrap decision looked like from the inside.

Dropbox's product history is a lesson in what happens when a simple, beloved consumer experience meets the pressure to become an enterprise company.

Facebook's business depends on knowing more about you than you'd consciously share. The product that generates this intelligence is the social graph itself. A case in data ethics, advertising architecture, and the PM's role in navigating them.

FedEx launched its website in 1994 — before most companies knew what "e-commerce" meant. How the world's leading logistics company used internet technology to build competitive advantages that took competitors years to match.

After 150+ food-tech startups folded in 2016, the Indian market was still standing — and more crowded than ever. A case on reading a competitive landscape before building your product.

How Foursquare pioneered location-aware discovery, then spent a decade trying to find the business that matched the technology it had already built.

Google started as a university project called Backrub. Its path from 1998 search engine to one of the five most valuable companies in the world is a study in how organisational structure enables product innovation at scale.

The AdWords auction didn't just monetize search — it redefined what "product innovation" means when the product is a marketplace between attention and intent.

How Harvey built legal AI that large law firms actually trust — the eval design, auditability, workflow integration, and buyer trust decisions that vertical AI requires when the cost of a wrong answer is a malpractice claim.

Four ex-Merrill Lynch executives built a financial data analytics platform in India in 2010 — before big data was a mainstream category. A case in data product design and analytics for PMs.

Netflix didn't win on technology or selection — it won by rethinking the business model, and then again by abandoning the business model that had just worked.

Instagram's journey from photo-sharing app to a $100B+ advertising platform is inseparable from the rise of the influencer economy. A case in product portfolio design at massive scale.

LinkedIn's 700M professional user base wanted live video. The product team wanted engagement. Trust and safety stood between both. A case in feature launch with high reputational stakes.

M&S had a complaints backlog and order processing problems. Their solution was ICOS — an integrated ordering system that unified online, phone, and in-store purchasing. A case in systems design and retail technology integration.

How Notion introduced AI writing assistance into a product where people store their most important thinking — and the onboarding, opt-in design, and data handling decisions that determined whether users accepted or rejected it.

PayPal processes hundreds of billions in payments annually. The product that makes this possible is invisible to most users — a real-time fraud detection engine built on machine learning. A case in data science as product strategy.

How Perplexity replaced a link list with a generated answer, and what the citation discipline, hallucination handling, and monetisation decisions looked like from a product perspective.

Pets.com had a real product insight and real demand — it simply tried to build a 2010 business in 1999, with 1999 infrastructure and 2010's cost structure.

How India's leading food-delivery platform navigated the pricing tension between platform take rate, restaurant margins, and consumer affordability — and what it reveals about marketplace pricing in practice.

In 2024 Replit shipped a multi-step agentic coding system that could plan, execute sub-tasks, handle failures mid-sequence, and stop itself before doing irreversible damage. A case in how orchestration choices — what runs in parallel, what gates on confirmation, how state passes between agents — are product decisions, not architecture footnotes.

ShareChat started in 2015 as a way to find shareable content for WhatsApp. By 2021, Google was reportedly in talks to acquire it for $1.03 billion. The path between those two points is a product roadmap study.

When a small US metals distributor launched a website in 2008 with no shopping cart, measuring its value was genuinely hard. A case in B2B website KPIs and the difference between traffic and business impact.

Starbucks had 17,000 stores in 49 countries and prices higher than the market average. Social media was how they turned a commodity product into a cultural identity — and kept customers coming back.

Tata Steel paid $12.15 billion for Corus in 2007, jumping from the 56th to the 6th largest steel producer in the world overnight. A case in M&A strategy, integration, and what a product manager's role looks like post-acquisition.

How a small team launched a product into a competitive discovery market using organic channels and community trust rather than paid acquisition — and what a senior PM takes from a constrained GTM.

TSI had seed funding, a smart bulk-purchase model, and a gap in the Broadway theatre ticket market. A case in applying product lifecycle thinking and SDLC methodology selection to a competitive market entry.

Topdox built a genuinely useful document collaboration platform out of Lisbon. It raised €1M, had real users, and shut down in 2015. The question it couldn't answer was what to measure.

Twitter's blue checkmark was meant to authenticate identity. It became a celebrity perk. When both meanings collided in 2017, the product team had no coherent answer — because the original design had never chosen one.

Urban Company survived COVID but emerged into a world where users' expectations had fundamentally shifted. Safety was no longer a differentiator — it was table stakes. A case in product-led trust recovery.

When Vedantu raised its Series A in 2017, the hardest question wasn't how to scale — it was what, exactly, the product was. A case in product portfolio clarity for edtech startups.