How These 3 Founders are building on Open Claw | E2248

Key Concepts

• OpenClaw: An open-source agentic platform (formerly Clawbot) allowing creation of personas/replicants to automate tasks.
• Agentic AI: AI systems capable of autonomous action and decision-making, beyond simple responses to prompts.
• Replicants: The term used for the AI agents created within OpenClaw.
• PostHog: An analytics platform used for tracking user behavior and data.
• Agentmail.to: A service providing AI agents with their own email address for task execution.
• Prompt Injection: A security vulnerability where malicious prompts manipulate an AI's behavior.
• Cron Job: A scheduled task that runs automatically at specified intervals.
• LLMs (Large Language Models): AI models like ChatGPT and Gemini, capable of understanding and generating human language.

The Rise of Agentic AI and OpenClaw: A Deep Dive

This episode of This Week in Startups focuses on the emerging landscape of agentic AI, specifically highlighting the open-source platform OpenClaw and its potential to revolutionize startup operations and beyond. The discussion centers around practical applications demonstrated by Pre Deshkumar (CEO of The Wellness Company) and early adopters Sean Louu and Vishnu, showcasing how OpenClaw is moving beyond theoretical potential to tangible results.

I. OpenClaw: From Viral Trend to Practical Tool

The conversation begins with Pre Deshkumar detailing his initial encounter with OpenClaw (formerly Clawbot) and its rapid adoption. He emphasizes the platform’s ability to create “replicants” – autonomous agents capable of performing complex tasks. He illustrates this with several examples:

• Automated Communication: A request from a team member (Jacob) via email triggered Awa (Pre’s OpenClaw replicant) to automatically create a document summarizing previous use cases and send it back. This process took approximately five minutes.
• Data Retrieval & Outreach: Awa was tasked with retrieving active user data for the Tempo product from PostHog, formatting it into a Google Sheet, and then sending personalized email requests to those users for reference purposes. This multi-step process, previously requiring significant manual effort, was completed rapidly.
• Email-Based Task Management: Utilizing Agentmail.to, Awa was given its own email address, enabling task delegation directly through email communication. A key security feature was implemented: Awa requires Pre’s approval before executing any task initiated by an external email. This mitigates the risk of prompt injection attacks.

II. The Power Shift: AI as an Employee

Jason Calacanis frames OpenClaw as a paradigm shift, moving AI from a “powerless” state (sandboxed LLMs) to a position of agency. He argues that OpenClaw effectively provides the capabilities of a mid-tier employee (estimated cost $70,000-$80,000 annually) at a fraction of the cost. Pre agrees, noting that tasks he previously delegated to a $30-$40/hour employee are now handled by his OpenClaw replicant. This highlights the potential for significant cost savings and increased efficiency.

III. Advanced Use Cases & Automation

The discussion expands to more sophisticated applications:

• Automated Content Creation: Awa was used to monitor news sources for relevant health and wellness articles. Upon identifying a study on longevity, Awa was instructed to create a blog post based on the research, significantly accelerating the content creation process. The next step, automating publication to the website, is also within reach.
• Visual Understanding & Integration (Sean Louu’s Work): Sean Louu demonstrated a system integrating OpenClaw with Meta Ray-Ban smart glasses and Gemini Live. This allows for real-time visual input and voice commands, enabling tasks like identifying products and adding them to an Amazon cart. The system utilizes Gemini Live for real-time perception and then delegates tasks to OpenClaw.
• Automated Customer Support & Bug Reporting (Pre Deshkumar): Awa monitors Pre’s inbox for support requests. When a bug report is received, Awa accesses the company’s GitHub repository, identifies the relevant code, proposes potential solutions, and posts the findings to a Slack channel for the development team. In one instance, Awa identified a bug, and another AI agent (Claude) fixed it and created a new branch for testing.
• Scheduled Information Gathering (Cron Jobs): Awa is programmed to run cron jobs that search for relevant news articles and deliver them to Pre, providing a curated information feed.

IV. Impact on Startup Economics & Future Trends

Jason Calacanis posits that OpenClaw and similar technologies will dramatically lower the cost of building and running startups. He predicts a surge in the number of apps created as the barrier to entry decreases. He suggests founders will be able to skip funding rounds, retaining more equity in their companies. Pre agrees, stating that AI tools have already enabled his company to ship multiple iOS apps with a smaller team.

V. Security Considerations & Mitigation

The discussion acknowledges the security risks associated with granting AI agents access to sensitive data and systems. Pre implemented a crucial safeguard: Awa requires explicit approval before executing any task initiated by an external source, mitigating the risk of prompt injection attacks. Vishnu highlighted the importance of running OpenClaw in a sandboxed environment (virtual machine) to protect sensitive data.

VI. Investment & Future Development

Vishnu, an engineer who built a hosted OpenClaw solution (agent37.com), impressed Jason Calacanis with his initiative and rapid progress. Calacanis offered Vishnu $125,000 in funding for a 7% stake in his company, contingent on due diligence, and invited him to join the Launch accelerator. Sean Louu also received a similar offer.

Data & Statistics

• Estimated cost of a mid-tier employee: $70,000 - $80,000 per year.
• Agent37.com hosting cost: Approximately $1.25 per month.
• OpenClaw instance hosting cost (via Vishnu): $7 - $8 per month (cheapest option found).
• Vishnu’s user base (as of the recording): 71 active users.

Conclusion

The episode paints a compelling picture of the transformative potential of agentic AI, particularly through the lens of OpenClaw. The demonstrated use cases highlight the ability to automate complex tasks, reduce operational costs, and empower individuals and small teams to achieve more with less. While security concerns remain, the proactive measures discussed demonstrate a growing awareness and commitment to responsible AI development. The investment offers extended to Vishnu and Sean Louu underscore the belief that this technology represents a significant opportunity for innovation and disruption. The core takeaway is that AI is no longer just a tool for information retrieval; it's becoming a proactive partner capable of driving business operations and accelerating innovation.