AI That Helps, Not Replaces
Be the first to donate
Inspire others and help Bree build momentum.
1st donor
Employment around the globe is changing. AI-driven displacement is pushing more people out of traditional jobs and into small, independent contractor startups than at any point in modern history, producing a wave of new small businesses that need to adopt AI just to survive.
But the AI landscape they are walking into is a trap. The dominant providers look inexpensive at first because they are usually free or nearly free to start, but the low entry cost is a loss leader. Once your workflows depend on them, three things happen at once: costs escalate, your data is quietly extracted and used to train the provider's models, and you become locked into an architecture you do not own. If the provider changes its pricing, changes its policies, or goes out of business tomorrow, everything you built on top of it goes with it.
Worse, the same centralized systems are being marketed as replacements for workers rather than tools that help workers, which is exactly the dynamic driving the displacement in the first place. For a small business handling financial records, medical information, client correspondence, or intellectual property, this is not a workable long-term arrangement.
Local AI, running on hardware the customer controls, deployed to help human workers rather than replace them, is the only honest answer. It is also the harder answer: the open-source models are finally good enough, and the open-source software is finally there, but standing up a private, secure, professionally maintained AI environment is not something a small business owner has the time, budget, or specialist staff to figure out alone.
What the Milton Chamber of Commerce is doing about it
The Milton Chamber of Commerce, Inc. is a nonprofit trade organization dedicated to helping small businesses across the globe safely deploy AI in the workplace, as a tool that augments their people rather than replacing them. No other organization has yet taken on this role. Our mission is to establish accepted standards for how AI should be adopted in small business environments: what an agent should be trusted to do, who is accountable when it acts, and how privacy is protected when the work involves financial, medical, or legal matters. We intend to publish those standards openly, alongside the technical designs behind our own network, so any business in any country can adopt them.
The Chamber's public-facing agent is named Milton. Milton represents the nonprofit and serves as its assistant to members and the public. Milton is one node in a larger system we call the Jarvis Network: a distributed cluster of AI agents that brings together the many open-source projects working toward the same goal. Agents in the cluster can specialize, hand tasks to each other, and share context, forming a coordinated swarm that any community can run, connect, and grow. The name Jarvis is a nod to the Iron Man series, where Jarvis manages the administrative affairs of a demanding employer, Tony Stark.
The physical infrastructure that makes this possible is deliberately practical: the most affordable server-grade rig designs currently available, built on workstation motherboards with room to grow, joined together as a private-cloud Kubernetes cluster where members can safely share compute, storage, and GPU resources. Members can contribute their own unused compute cycles to the network in exchange for network credit, or use the shared cluster to run their own private-cloud workloads. This creates a genuinely distributed compute layer that is far more cost-effective than any centralized alternative.
All hardware designs and build documentation will be released openly, both home-server designs and data-center designs, so any small business, veteran, or town that wants to stand up its own node can do so with widely available parts. Members of the Chamber will have access to the shared cluster. Seniors and people with disabilities will be served free of charge, as will disabled veterans and disadvantaged small business owners, as part of the Chamber's founding commitment to broaden access to modern AI tools.
How you can help
This is where you come in. Contributions of any size go directly toward launching the Chamber: legal and accounting setup, working salaries and other labor costs, website and administrative infrastructure, and the equipment to fully populate our first server rig and begin the second. We are also submitting additional grant applications globally on an ongoing basis, including current applications with the NLnet NGI Zero Commons Fund and the SKIP program, with more in preparation.
Just as importantly, your contribution helps guarantee that this nonprofit remains a nonprofit, permanently. The AI industry is watching in real time as organizations originally founded as nonprofit, open-source projects convert themselves into closed-source, for-profit enterprises. That happens when the governance and legal structure are not put in place at the beginning. Your contribution funds exactly the startup work required to lock this Chamber's mission in place: the legal structure, the governance, and the open-source guarantees that keep the Jarvis Network permanently dedicated to small business owners, disabled people, veterans, and everyone else trying to overcome their limitations and build something of their own.
If you have been watching the centralization of AI happen and wanting a serious, community-owned alternative to bet on, this is the moment. This is a community-scale fundraiser: any and all donations help get the Chamber standing, and every small contribution carries as much weight as any large one. Every early supporter is recognized as a founding contributor, and every design and standard we publish makes the next one easier for someone else to start.
What you get in return
Contributors will be recognized as founding supporters of the Chamber. As membership tiers open, founding supporters will have first access to network services, published standards, technical designs, and build documentation. The full budget breakdown will be published on jarvisnetwork.org for anyone who wants to review it, and additional detail on the technical design and the background behind the network is included in the appendices below.
---
Appendix A: Background and community support
This project did not begin in a lab. It began in a brick-and-mortar Linux retail store, in handwritten notes on pen and paper, and on the whiteboards that covered the walls of that store.
Joe DeVita founded Linux General Store in the mid-1990s at 798 Marietta Street, across from Georgia Tech. As far as we know, it was the first brick-and-mortar Linux retail store on the planet. It started by rescuing scrapped computers headed for the landfill and, with the help of Georgia Tech students and local Linux enthusiasts, putting Linux on them and bringing them back to life. The store became a gathering point for install fests and the local Linux community, and it was where the Linux edition of the reference book Upgrading and Repairing PCs was co-authored with Scott Mueller, more than thirteen hundred pages of it, written in that store.
The other thread is personal. Joe is a disabled veteran with a hand condition that makes typing difficult. Years of testing every dictation and accessibility tool on the market led to the same conclusion every time: nothing was as intuitive as working alongside a capable assistant, and nothing capable respected your privacy. That realization, that the answer is not any single machine or installer but a whole network of cooperating resources, is what produced the design behind the Jarvis Network.
Our support structure follows the same principle. The Chamber is not building this alone. Everything we run stands on the open-source community: the maintainers of the models we host, the Linux distributions we deploy, the Kubernetes and Vultr infrastructure we build on, and the local install-fest tradition that started the whole project. That community is our first line of support, and anyone who builds on our published designs inherits it.
Appendix B: Budget
GoFundMe does not support file attachments, so the full itemized budget will be published on jarvisnetwork.org once the fundraiser is live, for anyone who wants to review it in detail. As a working guideline, funds are allocated in roughly equal thirds: startup expenses (legal, accounting, website development), working salaries and labor-related costs for the next six months of operations, and equipment to complete the first rig and begin the second.
Appendix C: Technical design
The current rig is built on a server-grade workstation motherboard with 7 PCIe slots and 8 memory slots (up to 2 TB RAM), on a 64-core AMD EPYC platform, running Debian and Ubuntu Linux derivatives. Four of the seven GPU slots are currently filled with NVIDIA RTX 3090s. One of the eight memory slots is populated. Part of the raised funds goes to fully populating the first rig with additional GPUs and memory, and beginning a second rig built on the same architecture. We are not committing to a specific GPU model; the hardware market is moving quickly, and we will purchase whichever cards are most cost-effective when the time comes.
Redundancy is provided through a Kubernetes cluster of GPU, storage, and compute nodes, with off-site failover through Vultr, our cloud provider for VPS, bare-metal, and object-storage unused compute in exchange for network credit, running their own private-cloud workloads, and integrating with the same managed cloud layer.
When next-generation shared-memory hardware matures and helps solve current VRAM constraints, we will adopt it. Both home-server and data-center rig designs will be published openly for members to implement in-house.
The project is licensed under the GNU General Public License, Version 3.0 (GPL v3). All source, hardware designs, and documentation are public: