Tech people love new words.

Peer-to-peer. Mesh Networks. AI. Local-first. Decentralization. Data sovereignty. Solar punk. Regeneration. Sometimes these words point to something real. Sometimes they become stickers we put on the same old systems: more extraction, more centralization, more power for the people who already have too much of it.

This talk is about what happens when the buzzwords have to touch the ground.

For the past years, I have worked with Awana Digital alongside Indigenous and frontline communities using open-source, offline-first, peer-to-peer tools for land defense, cultural mapping, and environmental monitoring. One of these tools is CoMapeo, a peer-to-peer mapping platform used in places where internet is limited, unreliable, or simply not the center of life.

That last part matters. In a lot of tech spaces, lack of internet is treated as a problem to solve. In the territories where I work, the problem is often deeper: who controls the infrastructure? Where does the data go? Who can see it? Who can take it? Who gets to decide what should exist in the first place?

Peer-to-peer technology becomes interesting there not because it is elegant architecture, although it can be. It matters because sovereignty is real. Communities already understand sovereignty through land, language, governance, kinship, and struggle. The question is whether our digital tools respect that, or whether they quietly pull people back into clouds, platforms, and systems they cannot control.

From there, I want to talk about AI.

AI is usually presented as the future: bigger models, larger data centers, more data, more magic. But AI is also a new extraction machine. It extracts minerals, energy, labor, attention, language, and culture. For Indigenous languages, this is especially dangerous. We may soon see a world where a company’s model can speak a language that living communities have been violently pressured out of speaking themselves.

That is not a future I want to help build.

So what would another path look like?

The idea I want to share is simple: something like CoMapeo, but for language data. Peer-to-peer, offline-first, community-governed language data collection. People could record stories, conversations, songs, place names, teachings, or everyday speech. They could transcribe, translate, review quality, and decide together what can happen with that data. Maybe it trains a model. Maybe it stays only in the community. Maybe some things should never enter a machine at all.

From there we can imagine smaller, stranger, more useful AI systems. Local machines. Solar-powered infrastructure. Voice-first tools for oral cultures. Open-source pipelines governed by the people whose language, knowledge, and futures are at stake.

But this is not a talk about using AI to “save” Indigenous cultures. That framing is backwards. Indigenous peoples are not waiting for AI to bring intelligence, resilience, or climate solutions. Many of the most important technologies already exist: how to live in community, how to cultivate abundance without destroying the forest, how to govern land together, how to belong to a place, how to keep a planet alive.

Those are the real advanced technologies. For human and planetary wellbeing not for industrial production.

This session is for people building decentralized technology who want to ask harder questions than “how do we scale this?” Peer-to-peer is not just a protocol choice. AI is not just a product category. And a thriving planet will not come from better buzzwords.

It will come from tools humble enough to serve living worlds instead of replacing them.