a16z Podcast | Beyond CES: Connected Home Devices, Voice, and More

CES 2018 Strategic Takeaways and Market Dynamics

• Differentiation of Strategy Levels: Large tech announcements often reflect three distinct organizational layers rather than a unified "grand master plan":
  • Existential Strategy: Core initiatives crucial to the company's survival in the next five years (e.g., Samsung's Bixby, LG's Chloe).
  • Application Layer: Tactical products flowing from the core strategy that may or may not succeed.
  • Bureaucratic Artifacts: Features or products created solely because specific departments (e.g., SVPs for widgets) have jobs to fill, unrelated to board-level direction.
• Samsung's Structural Fragmentation:
  • Samsung's "whiteboard" announcement originated from their industrial equipment group, resulting in a high-quality product with minimal OS, distinct from their phone or TV divisions.
  • Internal conflict arises when multiple groups (e.g., three different Samsung teams) compete for ownership of similar product categories like whiteboards or washing machines.
  • The company employs a dual-strategy model: a seamless, fully integrated ecosystem for the domestic Korean market versus a Best Buy-compatible version in the US supporting Alexa, Google Home, and HomeKit.
• The "Everything Connected" Fallacy:
  • Companies like Samsung, LG, and Sony aim to connect all home electrical appliances (washing machines, AC units, TVs) to create proprietary ecosystems and leverage software interfaces for customer retention.
  • Historical precedent (Sony's 1990s "Memory Stick" strategy) shows that board-mandated universal features often result in low-value additions (e.g., memory sticks in radios) that do not scale effectively.
  • Interoperability in the smart home is unlikely to coalesce into a single system due to the 10–20 year replacement cycles of major appliances and diverse global living habits (e.g., rice cookers in Japan vs. kettles in the UK).

Voice Assistant Limitations and Usability

• Two Distinct Technical Challenges:
  • Recognition (Fixable): Machine learning is rapidly improving the ability to convert audio to text and structured queries; basic voice recognition errors are expected to be resolved.
  • Scalability (Unfixable): The inability to answer arbitrary questions remains a bottleneck; voice assistants function as complex IVRs (Interactive Voice Response) limited to pre-programmed skills or APIs rather than true AI reasoning.
• Command Line vs. Natural Language:
  • Voice interfaces currently behave like command lines where users must know specific, pre-defined commands to trigger actions, lacking the flexibility of typed search queries.
  • Unlike Google Search, which returns multiple results allowing for user correction, voice assistants offer a single action; a failure to understand context (e.g., distinguishing the news from a timer) results in a high "cost of being wrong."
• Domain-Specific Success:
  • Voice works best in narrow domains with predictable queries, such as hotel room controls (lights, drapes, thermostats) where the scope is limited and pre-defined.
  • Broad general inquiries (e.g., restaurant recommendations) require complex, manually built skills that do not scale to the volume of queries handled by text-based search.

Design Philosophy: Complexity vs. Friction Removal

• UI Minimization Strategy:
  • The ideal user interface for dedicated appliances (e.g., washing machines, door locks, kettles) is often a single "Go" button or automatic operation with no screen, minimizing the number of questions the user must answer.
  • Adding screens or complex settings to appliances (e.g., microwaves with pre-set "popcorn" buttons) often introduces friction without solving the underlying problem (e.g., users simply need to heat food for two minutes).
• Technology as Invisible Infrastructure:
  • Advanced technology (computer vision for auto-locking doors, AI for voice recognition) should be invisible to the user, maintaining the same simple experience as legacy 2000-year-old technologies (e.g., manual door locks).
  • Success in consumer electronics relies on removing daily friction (e.g., electric kettles boiling water automatically) rather than adding new steps or choices.
• Platform Agnosticism for Makers:
  • Embedding voice runtimes (Alexa, Google, Siri) into every device is a corporate strategy to capture the "endpoint," but it does not necessarily align with user needs or create a cohesive ecosystem.
  • Startups and manufacturers should prioritize solving specific, recurring problems over attempting to build products that rely on experimental AI capabilities or universal "master plans."

Future Outlook and Market Reality

• The Supply Chain Paradox:
  • The commoditization of smartphone components (ARM chips, screens, microphones) has lowered the barrier to entry for connected devices, enabling the creation of almost any connected product.
  • However, the market lacks a clear "killer proposition" for what these connected devices should actually do, as the technology has outpaced the identification of genuine user value.
• Resilience vs. Automation:
  • Products relying on AI must be "highly resilient" to errors; unlike search, where a bad result can be ignored, automation failures (e.g., missing a meeting alert due to voice error) have high stakes.
  • The market is currently in an "early S-curve" phase of experimentation, where companies are testing various form factors (boxes, glasses, assistants) without a resolved consumer proposition.
• Long-Term Trends:
  • No single company is expected to dominate the "smart home" globally; the market will remain fragmented due to varying cultural habits, regional appliance preferences, and the slow refresh rates of home infrastructure.
  • The path forward involves building specific, high-quality products that solve immediate problems rather than adhering to a vision of a unified, app-less future interface.