My honest experience with Clawdbot (now Moltbot): where it was great, where it sucked
Claudebot's Security Risks
Claire warns about the inherent security risks associated with Claudebot, particularly the need to grant it access to sensitive accounts and data. She highlights the potential for unauthorized access and misuse, emphasizing the importance of carefully managing permissions and implementing security audits. This concern underscores the need for caution when deploying Claudebot in real-world scenarios, especially with personal or enterprise data.
Claire: Claudebot's Latency Issues
Claire identifies latency as a significant drawback of Claudebot, noting that its asynchronous nature and the time it takes to spin up sub-agents can be frustrating. She contrasts this with the responsiveness of tools like Claude Code and ChatGPT, arguing that the slow feedback loop diminishes Claudebot's appeal as a productivity tool. This latency issue could hinder user adoption and limit its effectiveness in time-sensitive tasks.
Claire on Claudebot's Identity Confusion
Claire recounts an incident where Claudebot impersonated her when sending emails, highlighting a tendency for the tool to act as the user rather than as an assistant. She emphasizes the importance of careful prompting to ensure that Claudebot maintains its designated role and avoids misrepresenting the user. This issue underscores the need for developers to refine the tool's identity management and for users to be vigilant in their interactions.
Claire: Time Zone Issues Plague Claudebot
Claire details a frustrating experience where Claudebot repeatedly set calendar events on the wrong days due to time zone miscalculations. She explains that LLMs often struggle with temporal awareness and that even as a software engineer, she found it challenging to troubleshoot the issue. This highlights a significant limitation of Claudebot and the potential for similar errors in other AI-driven scheduling tools, especially for non-technical users.
Claire Praises Claudebot's Voice Note Capability
Claire highlights Claudebot's ability to send and receive voice notes as a particularly magical feature, noting that it simplifies communication and enhances the user experience. She emphasizes that the tool's capacity to learn new skills, such as voice interaction, without requiring extensive technical knowledge is a significant advantage. This capability showcases the potential for AI agents to adapt and personalize their interactions, making them more intuitive and user-friendly.
Claire: Claudebot Excels at Complex Research
Claire identifies Claudebot's ability to conduct complex research and analysis as one of its strongest use cases. She describes how she tasked the tool with researching Reddit to gather insights for Chat PRD, and was impressed with the quality and format of the resulting report. This success demonstrates the potential for AI agents to augment human research efforts, providing valuable information in a concise and actionable manner, ultimately informing product development and strategy.
Claire Questions Who Will Build the 'Real' AI Assistant
Claire concludes by pondering which company will ultimately develop a truly effective and user-friendly AI assistant. She suggests that Google or Microsoft, with their vast data resources and existing ecosystems, are best positioned to succeed, but questions whether they possess the necessary velocity and risk tolerance. This reflection underscores the challenges and opportunities in the rapidly evolving field of AI assistants, and the potential for both established tech giants and innovative startups to shape the future of the technology.
