What Duckbill changed

• The homepage headline changed from a technical focus ("The real world doesn’t have an API") to a consumer benefit ("You've never been so productive doing nothing").
• Added dozens of specific consumer use cases, such as negotiating cable bills, finding therapists, and getting security deposits back.
• Reorganized the site to prioritize three distinct audiences: individual users, AI developers, and "Doers" (human workers).
• Introduced a "Human API" section, emphasizing that they use real people to handle tasks that AI cannot complete, like sitting on hold.
• Added a testimonials section featuring quotes from founders and executives about personal productivity gains.
• Shifted the primary call-to-action from "Tell us your volume" (B2B) to "Get started" (B2C signup).

Duckbill has shifted its primary focus from a developer-facing API to a consumer-facing personal assistant app. They are now positioning themselves as a solution for "soul-crushing" life admin tasks like fighting insurance claims and canceling gym memberships, rather than just a backend tool for other AI agents.

• Added specific performance data showing that humans finish 70% of calls while AI alone only finishes 14%.
• Introduced a new three-step workflow section: Send the goal, Watch it run, and Get the outcome.
• Added a list of "real-world" failure points for AI, such as vendors only taking checks or offices having specific pickup hours.
• Updated the "Human involvement" section to a new "Human where it matters" section that emphasizes their hybrid architecture.
• Added an interactive task explorer showing 36 specific steps across different task types.

Duckbill is shifting its messaging to be more transparent about the limitations of AI. By highlighting that humans finish 7 out of 10 calls, they are positioning their hybrid "human + AI" architecture as a more reliable alternative to pure AI agents that often fail in the real world.

• Added a new FAQ section explaining the product, integration methods, and pricing.
• Changed the core value proposition from "execution layer" to "execution API for AI agents."
• Introduced a "pay per task" pricing model to replace vague enterprise descriptions.
• Added specific support for the Model Context Protocol (MCP) to integrate with Claude and ChatGPT.
• Shifted messaging to focus on reliability, claiming a 93% success rate for real-world tasks compared to 63% for AI alone.
• Added a comparison to Stripe, positioning the service as a way to avoid becoming a "services company."

Duckbill has shifted from a general service tool to a developer-focused "execution layer." They are positioning themselves as the API for the physical world, allowing AI developers to outsource difficult real-world tasks like phone calls and errands.