The HumanX conference in San Francisco this week saw Anthropic’s Claude emerge as the unspoken headline. Unlike the flashy product launches at rival events, attendees focused on how Claude 3.5 handled real-world tasks. Executives from Microsoft, Salesforce, and Shopify cited its lower latency and fewer hallucinations compared to competitors. One banking CTO said the model’s consistency in financial document analysis stood out during closed-door demos.
Anthropic’s booth drew queues for over two hours each day. Engineers tested Claude’s API against Google’s Gemini 2.0 and Meta’s Llama 3.1 in live coding challenges. Results showed Claude 3.5 completing Python debugging 23% faster with 18% fewer errors. A Stanford AI researcher noted the model’s self-correction features were already being integrated into internal tools.
The company’s closed-source approach also sparked debate. While OpenAI’s Sora and Google’s Veo showcased flashy generative media, Anthropic emphasized reliability over novelty. A Fortune 500 tech director said this shift mirrored how AWS disrupted cloud computing by prioritizing stability over early hype.
Anthropic co-founder Jared Kaplan avoided the main stage but held private sessions. Attendees described him as measured—focusing on safety benchmarks rather than sales pitches. The contrast with Elon Musk’s X.AI booth, which featured a $5,000 VR demo, was stark. One attendee quipped that Anthropic’s presence felt like IBM in the 1990s—unassuming but unavoidable.
By the final day, Claude’s API pricing became the most requested detail. Anthropic announced a 20% discount for enterprise contracts signed before June. The move underscored how practical performance now outweighs spectacle in enterprise AI adoption.
Source: techcrunch.com