Advocacy
Representation and policy engagement on rules that affect open weights, local inference, training data, compute, and consumer-facing AI.
Gives operators a voice before the rules are written for them.
Services
Infrastructure for the field.
Heiwig does not operate your models, your classroom, or your stack. It serves the people and organizations who build, teach, and run open-source AI: advocacy, standards, teaching, and a room to compare notes.
The customer is the member. The beneficiary is the field. Capability stays in public hands only if the operators have a place to stand.
Role
What this nonprofit does
It represents members before lawmakers and regulators. It writes and promotes ethical practice. It teaches the operational and legal craft. It convenes the field. It gives serious vendors a channel that is not a closed marketplace.
Nonprofit does not mean free, and it does not mean ownerless charity. Dues, event fees, training, and sponsorships — when they exist — fund the mission. They are not distributed as profit to shareholders. Heiwig will post membership terms when that program is real. We will not invent a dues schedule.
Offering
What members get
A service catalog for open-source AI operators — not a substitute for running the work yourself.
Compliance intelligence
Briefings on legal and regulatory change — including privacy, consumer protection, and open-source carve-outs such as those in the EU AI Act.
Reduces the cost of flying blind as the rules move.
Education
Cohorts, workshops, publications, and methods that produce operators — people who can run, measure, and teach open systems.
A right you cannot exercise is ceremonial.
Networking
Peer rooms for labs, teachers, veterans, vendors, and public-interest builders. Introductions, working groups, and shared problems.
Isolated operators reinvent the same stack and lose the same fights.
Standards and ethics
A public code of practice: open weights and methods, test-before-release, no monopoly dressed up as safety. See Ethics.
Trust is cheaper when the rules are written in the open.
Market visibility
A place for members and partners to be found — talks, notes, and honest listings. Not a vendor mall.
Good work should not depend on a closed procurement list.
Industry research
Surveys, benchmarks, and public notes on what open models, local compute, and teaching cohorts actually do.
Leaders should decide with evidence, not press releases.
The ethics and release standard lives on the Ethics page.
Questions
Why an operator joins
- What new rule affects our right to train, run, or publish a model?
- How are peer labs and classrooms using local open-source AI without renting the stack?
- Which tools and partners are actually open — weights, code, and methods — not just an API?
- How do we train people to operate systems instead of chatting with a landlord?
- How does a small lab or school get heard when policy is written by incumbents?
Who
The membership universe
- Independent labs and open-weight builders
- Teachers, schools, and veteran training programs
- Public-interest tooling and local-compute operators
- Consultants, researchers, and curriculum authors
- Vendors who ship open methods rather than permission slips
Plain language
Representation and teaching. Not delivery of the work.
The network, the briefings, the collective voice, and the teaching layer. Not your inference bill. Not a substitute for running the models yourself.