Vision

The problem

Cultural knowledge in South Africa is overwhelmingly oral, embodied and local. It lives in elders, healers, cooks, herders, weavers and storytellers; it is passed on by doing and by speaking. That chain is thinning:

• Urbanisation and migration separate generations from the places and people that hold the knowledge.
• Education and media centre a handful of dominant languages; many South African languages have a thin digital footprint.
• When a knowledge holder dies unasked, the knowledge usually dies with them.
• What is written down is scattered across academic papers, museum archives and private notebooks — often in English, often without the community that originated it ever seeing it again.

A young person today can identify a Pokémon faster than the plant growing at their grandmother's gate.

The idea

Reservoir is a free app and open platform that makes cultural knowledge findable, learnable and living again:

• Point a camera at a plant → learn what it is, what it is called in your language, how it has been used, how it is cooked or prepared, and what to be careful about.
• Record what you know — a name, a remedy, a recipe, a song — and have it safely attributed, preserved and shared on your terms.
• Explore how practices from the past fit into modern life.

Why an AI that is native to the mission

A generic chatbot would flatten, guess and sometimes fabricate. Reservoir's intelligence layer is designed around the mission from the ground up:

• It knows where its knowledge comes from and says so on every claim.
• It asks before it assumes — it can interview a knowledge holder and structure what they say, in their language, without putting words in their mouth.
• It respects consent: some knowledge is public, some is community-only, some is sacred and never leaves its custodians. The AI enforces that, it does not erode it.
• It is a fleet of specialised agents (identify, elicit, transcribe, translate, verify, cross-reference, safeguard) with humans holding the final authority.

Why South Africa first

South Africa has 12 official languages (including SASL), additional recognised Khoi and San languages, extraordinary floral diversity (the checklist of the national flora lists on the order of 21,000 species, over half endemic), a living tradition of plant-based medicine and food, and — importantly — a legal and policy framework for indigenous knowledge that is among the most developed in the world. It is a place where this project can be done properly and become a model for elsewhere in Africa and beyond.

Success looks like

Within the first years:

• A teenager in Thohoyandou, Soweto or the Eastern Cape photographs a plant and learns its name in their home language, spoken aloud by an elder.
• A grandmother records a recipe once, in her language, and her grandchildren can find it forever.
• A botanist and a traditional healer both trust the safety notes.
• A community can look at the project and say: "this is ours, and we control it."

Failure looks like

(The anti-goals — things that would make Reservoir a failure even if the software worked perfectly.)

• Extraction: knowledge harvested from communities and monetised or published without their consent or benefit.
• Hallucinated tradition: AI producing plausible-sounding but false cultural or medicinal claims.
• Harm: someone is poisoned by a mis-identification, or an endangered plant is stripped because the app revealed where it grows.
• Flattening: a handful of languages and a "generic African" culture, rather than the real diversity.
• Exclusion: an app that only works for people with new phones, fast data and English.
• Abandonment: a launch with no plan for maintenance, moderation and stewardship.

Non-goals

• Reservoir is not a medical product, a diagnosis tool or a place to buy remedies.
• It is not a social network.
• It does not try to "digitise everything"; some things are better left oral, and the community decides which.
• It is not a replacement for elders and teachers. It is a bridge back to them.