You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
Memoria Viva Ngäbe is an open-source project focused on building the first binational framework for Indigenous data sovereignty and recognizing diverse knowledge systems in AI. Backed by Hemispher-IA Business & Knowledge Hub, we’re creating an offline-first, mesh-network platform—adapting Colombia’s Recordari design—to help Ngäbe-Buglé communities along the Panama–Costa Rica border document, manage, and control their cultural data. The system works even without the internet, putting communities in charge and keeping their information off external servers. For 18 months, we’re piloting this across 10 to 12 communities, directly hiring six young Ngäbe “Digital Weavers,” and developing modular, open tech that anyone can reuse. Right now, Indigenous knowledge is almost invisible in AI training data, and Indigenous areas are wide open to digital exploitation. We’re changing that.
Our main goals include:
1. Indigenous Data Sovereignty Infrastructure: We’re building and open-sourcing a technical and governance model so Indigenous communities can classify and audit their data—Public, Restricted, or Sacred—on their terms. This directly counters AI data exploitation.
2. Offline-First Collaborative Systems: We’re turning battle-tested memory architectures into a peer-to-peer app that still works when you’re offline. This can serve any community with poor connectivity, whether that’s rural villages or disaster zones.
3. Epistemic Diversity in AI & Replicability: We’re fine-tuning language models for Ngäbere (spoken by about 200,000 people but barely present online) and creating open datasets. Once tested, we’ll release a playbook so other cross-border Indigenous groups—like the Maya, Guaraní, or Asháninka—can copy the model at almost no additional cost.
Our methods are a blend of deep academic rigour (working with the Recordari creators and Universidad del Rosario, Colombia), trustworthy ethical checks (Latin American Council on Ethics in Technology -CLET-, Center for AI & Digital Policy -CAIDP-, and, most importantly, the leadership and paid work of local Indigenous talent who co-design and run the tech.
We’re asking for $150,000 to fund the project for 18 months. Here’s how we’ll spend it: - 68% ($102,000) on people. Everyone gets proper pay, not “volunteer” exploitation. This covers executive direction, top-tier tech, local field leads, and salaries for the Digital Weavers (at $450/month plus their tech kits). - 15% ($22,500) builds the tech—from local servers and the offline-first app, to tools for language model training. - 7% ($10,500) buys field gear—refurbished smartphones, solar chargers, and mesh antennas for real-world setup. - 10% ($15,000) handles legal advice, ethics audits, and operations across borders.
Hemispher-IA is a Panama-based innovation hub specializing in managing real- world, scalable impact projects across Latin America and the Caribbean. We mix software engineering chops, Indigenous governance know-how, and responsible AI research. - Executive Direction: Wolfgang Friedl leads, bringing three decades of experience in global partnerships and social innovation (with UNICEF, WHO, and more). - Technical and Methods: Arthur Picerna and Luisa María Rincón from Universidad del Rosario, the people behind Recordari. They’ve shaped the core architecture and oral history techniques. - Local Operators: OLINABP (Panama) and FAIR Costa Rica, both Indigenous-led AI groups with deep roots in these communities and proven records in local governance. - Oversight: CLET (ethics and certification), CAIDP (human rights/privacy), GCRAI (AI alignment), and Silicon Valley Certification Hub (AI standards).
Our track record? We’ve already deployed robust data systems in tough, low-connectivity environments.
Our community-centered governance frameworks aren’t stuck in theory—they’re tried, tested, and ready to scale.
If things go wrong, why, and what will be left? Honestly, integrating Indigenous knowledge fully into global AI is a long game, not something we’ll finish in 18 months. We might hit big challenges—legal issues between Panama and Costa Rica, connectivity problems, or bureaucratic slowdowns. But even if we hit roadblocks, your support means we’ll still deliver:
1. A tested, open CARE Protocol that any Indigenous group can use to manage their data.
2. An offline app and mesh networking system, reusable wherever connectivity is low.
3. New NLP datasets and training tools for Ngäbere, giving other language communities a blueprint for their own tools.
4. A Binational Replication Playbook detailing all the hard lessons— technical, social, and ethical—so others can build faster. No matter what, these will remain as open, lasting resources for the global AI, climate, and digital rights communities.
Over the past year, we’ve brought in strong in-kind support: technical mentorship from the Recordari team, free ethics auditing from CLET, and access to trusted local networks through OLINABP (Ngäbe-Buglé Future Leaders Organization (Panama) and (Foundation for Responsible AI) FAIR, Costa Rica. We’re talking with potential institutional funders like Senacyt (Panama) and KAS-PPI about co-funding the next phases.
This Manifund grant would be our first real cash injection to move from planning into building, hiring, and launching in the field.