bash-3.2$ ./profile

Harry Mahon

$ whoami

I’ve worked with early-stage startups and large enterprises, building from zero to one. Outside of work I make time for cooking, reading, travel, and the NBA.

$ cat experience.log

  • Startups: helped an early voice-agent company go from zero to category-leading. Worked across product and go-to-market; landed first customers. Lessons learned - fast, abundant inference, and empathetic design.
  • Enterprise: developed AI strategy, workflow adoption, and partner ecosystems.
  • Advisory: here through growth. Happy to make warm intros for win-wins. Reach out!

$ ls links

$ status

← Back to writing

cat writing/the-year-of-the-claw

About The Year of The Claw

Harry Mahon ·

The Claw is the law.

OpenClaw’s Origin

The early 2025 release of Claude Code sparked an explosion of autonomous coding agents. By the end of 2025, organizations like Shopify reported that some of their best engineers hadn’t written a line of code in months. Around the same time, OpenClaw, a fully autonomous agent capable of completing non-coding tasks, was released.

OpenClaw integrates with platforms like Google Workspace to complete tasks on your behalf, while retaining and learning from memory. The “Claw Craze” has led builders to automate sales workflows, clean inboxes, and manage entire workdays. Peter Steinberger, the creator of OpenClaw, envisioned an agent designed to get more done with full computer access and unlimited tool use. The concept has inspired a new wave of products, including Perplexity Computer and several open-source alternatives that perform tasks autonomously.

OpenClaw can also be self-hosted, making it one of the first agents to feel truly custom, able to integrate deeply with a user’s individual context. You can access your Claw through multiple channels, including mobile apps like WhatsApp and Telegram, which gives you flexibility over when, where, and what your agent can access. OpenClaw is pushing both businesses and consumers to rethink their workflows from first principles, spanning hardware, model choice, and data exposure.

NemoClaw and Hermes Agent

While users could self-host, OpenClaw’s rapid rise raised questions about its security, autonomy, and architecture. The agent’s exposure to personal and enterprise data sparked understandable concerns. To address these and make the technology enterprise-friendly, NVIDIA released NemoClaw during GTC 2026, offering a more controlled deployment environment.

In parallel, Nous Research launched Hermes Agent for the open-source community, attracting users who prefer more flexible and experimental systems. Today, the ecosystem of OpenClaw alternatives has formed two clear paths: enterprise teams turn to controlled stacks like NemoClaw, while startups, hobbyists, and open-source users embrace options like Hermes Agent.

Fitting into Business and Consumer Workflows

Claws are already replacing hours of manual work in business workflows. Tasks that steal time from core responsibilities, like sales teams entering data instead of engaging customers, can be fully offloaded. For example, a Claw can gather context before a meeting, summarize the conversation, schedule follow-ups, and track action items. In the future, enterprise use cases may expand to data-heavy workflows across product, cybersecurity, and operations, where maintaining contextual awareness is valuable.

On the consumer side, workflows are still emerging. Popular use cases include system maintenance, consolidating communication across fragmented channels, and cleaning long-neglected inboxes. These are practical but narrower than business applications. Many individuals describe their Claws as natural extensions of their “second brains,” augmenting existing systems like Notion, Obsidian, and Google Docs. Many people designing their claws around their workflows are noticing an explosion in their inference consumption, leading many to weigh the costs of buying hardware to support local, open-weights models.

Open-Weights Models & Hardware Investment

As Claws become part of daily workflows, token costs become noticeable. Increased usage creates a strong incentive to run inference locally for both cost efficiency and security. This shift is fueling demand for accelerated personal computing.

The Mac Mini has seen a resurgence as an affordable entry point. NVIDIA’s DGX Spark, once lacking clear product-market fit, has reemerged as a strong option for running open-weight models like the Qwen family. For high-intensity users, upfront hardware investment can quickly offset recurring token costs. At the same time, interest in ecosystems like Apple MLX is growing as the community optimizes models for local deployment on Apple Silicon.

OpenClaw has expanded the possibilities of AI agents far beyond coding, gaining traction even among non-developers. Remember, the best way to get familiar with these Claws is to start using one!

Resources

Thank you to Jan Zheng and Leily Moazami for their reviews. ❤️