GMP would operate across four levels of support — a way to clarify what kind of help a nonprofit actually needs, not a rigid menu of services.
1. Needs assessment. Many organizations do not initially need a custom AI system. They need help understanding where operational friction exists and whether AI is an appropriate intervention at all. This means interviewing staff, mapping workflows, and identifying where time is being lost. Sometimes the right answer does not involve AI.
2. AI setup. Helping organizations use existing tools safely and effectively — selecting the right products, configuring them well, and training staff to rely on them appropriately. Many nonprofits will never need a custom build. A clean, well-governed setup using existing tools may produce meaningful improvement on its own.
3. Workflow automation. Turning repeated work into durable workflows — intake summarization, multilingual communication, recurring reporting, and similar operational patterns. The test is whether the workflow makes the organization meaningfully easier to operate. A technically impressive workflow that saves no time, creates maintenance burden, or increases risk should not be considered successful.
4. Custom build (Good Model Labs). For cases where existing tools are not enough. Labs would build reusable tools from problems that appear repeatedly across deployments. Labs should not chase novelty. Its best work comes from repetition — seeing the same problem across several organizations, understanding it well, and building something that reduces the cost of solving it again.
<aside> 💡
Good Model Labs is the software and research arm — built to turn what the organization learns from deployments into shared infrastructure for the broader nonprofit ecosystem.
</aside>
GMP could eventually organize around major domains:
… and others. The purpose would be to develop domain expertise, shared project patterns, governance standards, and mentor networks suited to each setting.
A health project should have different standards than a food pantry project. A legal aid project should have different safeguards than an animal shelter project. Domain focus allows the organization to become sharper over time rather than spread evenly thin.