Built an AI Test Kitchen for Responsible NGO Experimentation
Client Earth
Designed and facilitated an AI Test Kitchen with ClientEarth to help legal, policy, campaign, communications, fundraising, and operations teams explore AI through hands-on, values-aligned experimentation.
Category / Type
AI strategy; responsible AI adoption; prototype sprint; workflow automation; internal experimentation lab; open-source AI infrastructure.
Client / Venture / Internal
Client work via Traverse in Space for ClientEarth.
Context
ClientEarth is an environmental law organization working across borders, systems, and sectors to protect life on earth. The organization was exploring how AI could support its legal, policy, campaign, communications, fundraising, and operations teams without compromising its values around responsibility, privacy, environmental impact, and human agency.
Lucas, with the help of Manel Heredero, shaped and facilitated an AI Test Kitchen: a structured experimentation environment where internal participants could identify high-value use cases, prototype practical tools, evaluate risks, and build AI literacy through direct experience.
Wicked Problem
For ClientEarth, AI adoption was not just a question of tooling. It was a behavioral, ethical, and organizational challenge.
The opportunity was clear: AI could help teams monitor signals, manage complex projects, support research workflows, and reduce operational drag. But the risks were equally real. ClientEarth’s work sits inside legal, advocacy, climate, and policy contexts where trust, confidentiality, provenance, and environmental impact matter. A generic “move fast with AI” program would have been misaligned with the organization’s culture and responsibilities.
The harder problem was helping thoughtful, skeptical employees become empowered participants in AI adoption without asking them to suspend their concerns. The Test Kitchen needed to create literacy, not hype; useful prototypes, not theater; and a path toward AI tooling that could respect ClientEarth’s values.
Solution
Traverse in Space helped create the AI Test Kitchen as a hands-on experimentation container: part internal lab, part strategy workshop, part prototype sprint, part AI literacy program.
The work focused on identifying “plates” — practical prototype candidates that could be tested with internal teams. Early plates included a signal crawler and scraper to monitor external developments, a project management tool to improve coordination and visibility, and a confidential note-taking workflow designed around sensitive information.
A central design choice was to explore fully local, open-source AI infrastructure where appropriate. This gave ClientEarth a way to investigate AI tooling while staying aligned with concerns around privacy, ethics, dependency, and ecological impact. Rather than treating those concerns as blockers, the Test Kitchen incorporated them into the design brief.
Lucas’s role combined facilitation and building: holding the experimentation container, translating AI concepts into organizationally relevant choices, and prototyping tools that teams could evaluate through use. Manel contributed to the facilitation, program design, and organizational framing of the Test Kitchen.
Outcome / Impact
The work is currently in progress.
The AI Test Kitchen has established a structured environment for ClientEarth to move from abstract AI debate into practical experimentation. It is helping the organization identify high-value use cases, build internal literacy, evaluate risks, and test prototype workflows before committing to broader implementation.
Expected outputs include one to three prioritized prototypes, evidence from testing, recommendations for future implementation, and a repeatable model for responsible AI experimentation inside the organization.
Role / Contribution
AI Test Kitchen facilitation, strategy, and prototype building.
Lucas shaped the experimentation container, supported AI literacy, helped translate organizational needs into prototype candidates, and built early technical workflows for testing. Manel supported facilitation, program design, and the structure of the experimentation process as part of Traverse in Space.
Supporting Proof / Artifacts
AI Test Kitchen proposal and operating model.
Structured four-month experimentation cycle.
Prototype candidates including signal crawler/scraper, project management tool, and confidential note-taking workflow.
Local/open-source AI exploration aligned with privacy, ethics, and ecological concerns.
Internal use-case discovery across legal, policy, campaign, communications, fundraising, and operations teams.