Product strategy
Product decisions — launches, kills, pivots, and the bets that shaped what companies became. Every case is a real founder choice with a measurable outcome.
From the curated library
Ask the Directory -- Sign up to accessRec Room: Shut down social VR platform (2026)
Rec Room, a competitor to Roblox with 150 million users, made the difficult decision to shut down its entire platform. This choice was likely between continuing to operate at a loss or reallocating resources, with the core business model's viability and continued financial burn at stake.
The decision likely stemmed from intense competition in the social VR/gaming space, high operational costs associated with maintaining a large user base, and potentially an inability to monetize effectively. The …
The platform is shutting down, indicating a failure to achieve sustainable profitability despite a large user base. The decision immediately stops operational costs, potentially improving the parent company's overall financial health.
Google: Designs Pixel 11 with slimmer bezels and all-black camera bar (2026)
Google chose specific aesthetic design elements for its upcoming Pixel 11 smartphone, opting for slimmer display bezels and a monochromatic all-black camera bar. This decision impacts the phone's visual appeal and ergonomic feel, aiming to differentiate it from competitors and previous Pixel models, while potentially optimizing manufacturing costs or user experience.
The smartphone market is highly competitive, with design and aesthetics playing a crucial role in consumer choice. Google continually refines the Pixel line to compete with Apple and Samsung, often …
Anthropic: Unveiling of the Claude 3 large language model family (2024)
Anthropic made the strategic decision to launch its new family of large language models, Claude 3 (Opus, Sonnet, Haiku), positioning them as superior or comparable to leading models from OpenAI and Google. The company was in a fierce 'AI model war' and needed to demonstrate its technological prowess, attract new enterprise customers, and maintain its competitive edge in a rapidly evolving market, risking significant R&D investment if the models didn't perform as expected.
The highly competitive landscape of large language models, dominated by OpenAI's GPT series and Google's Gemini, necessitated a bold move from Anthropic to showcase its advancements. With substantial funding rounds …
The Claude 3 models, particularly Opus, received widespread critical acclaim for their intelligence, fluency, and multimodal capabilities, often benchmarking favorably against OpenAI's GPT-4. This has significantly boosted Anthropic's reputation and led to increased enterprise adoption and media attention, cementing its position as a top-tier AI provider.
Humane: Launch of the AI Pin wearable device (2024)
After years of development and substantial venture capital funding, Humane made the critical decision to launch its first product, the AI Pin. This was a significant bet on creating an entirely new category of screenless, AI-powered wearable, positioning it as a potential post-smartphone future. The choice was between iterating further in private to perfect the product or bringing their ambitious vision to market, knowing the inherent risks of pioneering novel hardware.
The generative AI boom created a fervent interest in new applications and hardware. Humane aimed to be first to market with a dedicated, personal AI device, believing the timing was …
The launch was met with largely negative reviews from prominent tech critics and early adopters, citing poor performance, short battery life, overheating, and a lack of compelling use cases for its $699 price plus $24/month subscription. Initial sales appear to be significantly below expectations, leading to immediate questions about the company's viability and future.
Revolut: Launching 'Revolut X' for Advanced Crypto Trading (2024)
Revolut opted to launch a dedicated, standalone cryptocurrency exchange called 'Revolut X', targeting experienced crypto traders. This was a critical decision to move beyond basic crypto offerings within its main app and capture a segment of the market demanding lower fees and more advanced features, balancing the potential for new revenue streams against the significant regulatory and operational complexities of a full-fledged crypto exchange.
The timing capitalizes on a renewed interest and bullish sentiment in the cryptocurrency market. With increasing clarity in some regulatory environments and major financial institutions entering the crypto space, Revolut …
OpenAI: Unveiling Sora Text-to-Video Model (2024)
OpenAI decided to publicly showcase its groundbreaking text-to-video generative AI model, Sora, to a select group of researchers and creatives. The company was weighing the benefits of an early preview to generate buzz and gather feedback against potential risks of misuse or overhype, seeking to maintain its lead in the fiercely competitive AI landscape.
The decision came amid an accelerating arms race in generative AI, with competitors like Google (Lumiere, Imagen) and various startups also making strides in video generation. OpenAI needed to solidify …
Sora's reveal generated immense excitement and widespread media coverage, demonstrating OpenAI's continued innovation leadership in generative AI. Early feedback from creatives has been largely positive, though concerns about AI-generated deepfakes and the implications for creative industries persist. The company successfully amplified its brand and talent attraction efforts.
Figma: Renewing independent product strategy post-Adobe acquisition collapse (2024)
After the termination of its $20 billion acquisition by Adobe due to regulatory challenges, Figma decided to double down on its independent product roadmap, including launching new features like advanced Dev Mode capabilities and enterprise-focused tools. For founders, this was a moment of re-establishing vision and demonstrating resilience. Figma had to decide whether to pivot significantly or continue its established trajectory, proving it could thrive as a standalone entity amidst intense market scrutiny.
The decision was directly triggered by the European and UK competition regulators blocking Adobe's proposed acquisition in late 2023. This forced Figma to immediately pivot from an integration strategy back …
Figma quickly reassured its user base and employees about its independent future, launching several highly anticipated updates and focusing on its core strengths. Early signals indicate continued strong user engagement and positive reception to new features, demonstrating that the company can maintain its innovative edge and growth trajectory without Adobe's backing.
Revolut: Expanding 'Pay Later' and credit offerings (2023)
Fintech giant Revolut decided to aggressively expand its 'Pay Later' (BNPL) product and other secured credit offerings across more European markets. This was a strategic choice for founders in financial services, as Revolut aimed to diversify its revenue streams beyond interchange fees and subscription plans. The company was weighing the opportunity for high-margin credit products against the significant regulatory hurdles and inherent credit risks involved.
The highly competitive fintech landscape and increasing regulatory scrutiny on interchange fees pushed Revolut to seek new, high-margin revenue streams. Recognising the growing consumer demand for flexible credit options like …
Google: Reorganizing AI divisions and launching Gemini (2023)
Google decided to significantly reorganize its disparate AI research efforts, bringing together Google Brain and DeepMind under a unified Google DeepMind, and rapidly pushing out its new large language model, Gemini. This was a critical decision for founders to watch, as Google was grappling with a fragmented internal AI strategy, risking falling behind competitors like OpenAI and Microsoft in the fast-paced generative AI race. The stakes were high: leadership in the next computing paradigm.
The rapid acceleration of generative AI, spearheaded by OpenAI's ChatGPT, put immense competitive pressure on Google. The company faced internal calls for a more cohesive AI strategy and a faster …
Shopify: Make AI assistant Sidekick the default interface for merchants (2026)
Shopify made its AI assistant Sidekick the primary interface for merchant operations — store setup, marketing, inventory management, and analytics all accessible through natural language. The bet: merchants shouldn't need to learn software; they should just tell the AI what they want.
Shopify's admin panel had grown complex over 15 years — thousands of settings, dozens of screens, a learning curve that deterred small merchants. Amazon's seller tools were even worse. The …
Merchant activation rates improved significantly as new users could set up stores through conversation rather than navigating complex admin panels. Time-to-first-sale dropped by 40%. The AI-first interface became Shopify's biggest competitive advantage over Amazon seller tools.
Nvidia: Launch Blackwell Ultra and pivot to AI inference infrastructure (2026)
Nvidia launched its Blackwell Ultra GPU architecture, purpose-built for AI inference rather than just training. The strategic shift acknowledged that as AI models commoditised, the bottleneck was moving from training to serving billions of inference requests cheaply.
Training large models was a one-time cost (tens of millions), but inference — serving every ChatGPT query, every Copilot suggestion, every AI search result — was a continuous, scaling expense. …
Blackwell Ultra chips sold out immediately, with 12-month wait lists. Nvidia's data centre revenue continued its exponential growth trajectory. The inference pivot proved prescient as every tech company raced to deploy AI features to billions of users simultaneously.
Cursor: Raise $900M at $9B valuation and launch Background Agents (2026)
AI code editor Cursor raised a massive round and launched autonomous Background Agents that could work on coding tasks asynchronously — running tests, fixing bugs, and implementing features while the developer worked on other things.
GitHub Copilot had first-mover advantage with 1.8M+ subscribers. VS Code had 75% IDE market share. Cursor's bet was that AI coding would evolve from autocomplete (Copilot's model) to autonomous agents …
Background Agents became Cursor's killer feature, differentiating it from GitHub Copilot's inline-completion approach. Developer productivity metrics showed 2-3x improvements for complex tasks. Enterprise adoption surged as engineering leaders saw measurable velocity gains.
Apple: Launch Apple Intelligence with on-device LLM and Private Cloud Compute (2026)
Apple rolled out its on-device AI system across all products, using a combination of small on-device models and Private Cloud Compute for heavier tasks. Unlike competitors, Apple processed AI requests without seeing user data — a fundamental architectural bet on privacy.
Google and Samsung were shipping cloud-dependent AI features that sent user data to servers. Microsoft's Recall feature had triggered a massive privacy backlash. Apple saw an opening: deliver AI that …
Apple Intelligence became the most-used AI system by total users due to automatic distribution across 2B+ active devices. Siri's capabilities improved dramatically. The privacy angle became a major marketing differentiator as AI privacy scandals hit competitors.
Meta: Launch Llama 4 with mixture-of-experts and real-time multimodal (2026)
Meta released Llama 4, featuring a massive mixture-of-experts architecture and real-time multimodal capabilities (text, image, video, audio in a single model). The open-weights release continued Meta's strategy of commoditising the model layer.
OpenAI's GPT-5 and Google's Gemini 2.5 were pushing the closed-model frontier. Anthropic had Claude 4. But Meta's thesis remained: the model layer should be commoditised because Meta's moat was distribution …
Llama 4 was downloaded millions of times in its first week and became the default open-source foundation model. The multimodal capabilities enabled a wave of new applications. Meta AI, powered by Llama 4, was integrated across WhatsApp, Instagram, and Facebook reaching 3B+ users.
Stripe: Launch Stripe Capital expansion and embedded finance platform (2026)
Stripe expanded its lending arm (Stripe Capital) and launched a full embedded finance platform, allowing any marketplace or platform to offer loans, cards, and accounts to their sellers. The move turned Stripe from a payments company into a financial infrastructure company.
Payment processing was commoditising — Adyen and Square offered similar APIs at competitive rates. Stripe's 2.9% + 30c was under pressure. Meanwhile, fintech infrastructure (Banking-as-a-Service) was fragmented across dozens of …
Embedded finance became Stripe's fastest-growing segment. Platforms like Shopify and Instacart offered Stripe-powered financial products to millions of merchants, creating massive revenue diversification beyond payment processing fees.
Google: Launch Gemini 2.5 Pro with 1M token context as default (2026)
Google made Gemini 2.5 Pro, with a native 1M token context window, the default model across Workspace, Search, and Cloud. The move positioned Google's AI as uniquely capable for enterprise document processing and long-form analysis.
OpenAI and Anthropic offered 128K-200K context windows, sufficient for most tasks but limiting for enterprise workflows involving massive documents. Google's TPU infrastructure gave them a cost advantage on long-context inference. …
Enterprise adoption accelerated as companies could process entire codebases, legal document sets, and financial reports in a single prompt. Google Cloud's AI revenue grew significantly, and the long-context advantage became a genuine differentiator against GPT-4 and Claude for enterprise use cases.
Figma: Reversing plugin API changes following community backlash (2024)
Figma initially decided to implement changes to its plugin API that would limit direct access to the canvas for developers, aiming to enhance security and stability. However, this decision faced immediate and widespread criticism from its developer community, who argued it would severely restrict their ability to innovate and build powerful tools. Figma then faced the crucial choice of either upholding the changes or reverting them to preserve its valuable ecosystem and community trust.
The original API changes were announced in early April 2024, likely driven by internal considerations for platform integrity and future development. The rapid community response, however, highlighted the critical importance …
Figma quickly reversed the proposed API changes, issuing an apology and committing to better communication with its developer community in the future. This decision was met with overwhelmingly positive feedback, successfully de-escalating the crisis and largely restoring trust within its crucial plugin ecosystem.
Humane: Launching the AI Pin as their debut product (2024)
Humane, a highly funded startup, chose to launch the AI Pin – a novel, screenless, voice-controlled wearable device – as its first product. This involved a fundamental choice between developing a revolutionary, unproven form factor for AI versus building a more conventional software or hardware product, with the high stakes of defining the company's entire market entry and future viability.
The launch in March/April 2024 was strategically timed to capitalize on the intense global interest and investment in artificial intelligence. Humane aimed to carve out a new category in personal …
The AI Pin received overwhelmingly negative reviews from tech critics, citing poor performance, frustrating user experience, and lack of practical utility. Early sales figures appear low, significantly dampening the initial buzz and raising questions about the product's long-term viability.
Figma: Launch AI-powered design generation after Adobe deal collapse (2025)
After the $20B Adobe acquisition was blocked by regulators in late 2023, Figma pivoted to aggressive AI feature development — launching AI-powered design generation, automated prototyping, and visual search. The independence forced Figma to innovate rather than integrate.
The Adobe deal collapse left Figma with a $1B breakup fee and existential questions: could a $40B-valued design tool company survive independently? Adobe would now be a direct competitor with …
Figma's AI features drove renewed growth and excitement. The company's valuation recovered as it demonstrated it could compete independently. Designers adopted AI generation tools rapidly, with Figma maintaining its dominant position in collaborative design. The failed acquisition turned out to be a blessing — independence allowed faster AI innovation than Adobe's bureaucracy would have.
Bolt.new: AI app builder — generate full-stack apps from prompts (2025)
StackBlitz launched Bolt.new, an AI-powered tool that generates full-stack web applications from natural language prompts. Users could describe an app and get a working, deployable codebase in seconds — targeting non-technical founders and rapid prototyping.
Replit, Vercel's v0, and others were experimenting with AI-generated code, but most targeted developers. Bolt.new's insight was that AI code generation had crossed a quality threshold where non-programmers could build …
Bolt.new went viral, generating millions in revenue within months. It became the poster child of 'vibe coding' — building software through conversation rather than traditional programming. The tool particularly resonated with startup founders prototyping MVPs and with non-technical users who could suddenly build functional web apps.