What the platform actually does.
Nine pieces — sensors, certificates, language tools, payments, credits, insurance, reporting, compliance and hosting — running together from one place in Johannesburg.
A reading from each tree, every few minutes.
Small LoRaWAN devices clipped to monitored trees send back sap-flow, soil moisture and canopy temperature, plus their own battery level. The readings land in our Johannesburg database in seconds, even from areas with weak mobile signal.
- Sap-flow, soil moisture, canopy temperature and battery on each reading
- Tree positions stored with their plot, so you can roll up by site
- Unusual readings get flagged in the dashboard before they become problems
- Raw readings kept for 30 days; the summarised history kept indefinitely
Each lot carries its own proof of origin.
When a harvest is declared, every healer who contributed signs off on their share. The lot then gets a certificate that records who picked what, when and where. Anyone with the QR can check the certificate themselves — no app, no account, no need to take our word for it.
- A signed entry per contributor on every lot
- Tampering with the record afterwards breaks the certificate
- Verification works offline from the QR on the pack
- Buyers, auditors and regulators can all check the same certificate
Voice notes in isiZulu, isiXhosa, Sesotho or Ndebele.
A healer can send a voice note in the language they prefer. We transcribe it, store it, and reply in the same language. The models we use to do this run on our own servers in Johannesburg, so the audio never leaves the country.
- Voice transcription in four Southern African languages
- Search and retrieval that works across the contributors' words
- Used to compute royalties when a contribution is referenced
- No third-party AI provider involved at any point
When their knowledge gets used, they get paid.
Each time a corporate query draws on a healer's contribution, a small royalty is recorded against their account. Once a day, balances above the payout threshold are sent to MTN MoMo. The healer receives a WhatsApp statement showing every line item.
- Royalty calculated per citation, weighted by relevance
- Records are tamper-evident — they cannot be quietly edited
- Payouts via MTN MoMo, retried automatically on failure
- WhatsApp statements in the language the healer chose
Numbers that tie back to actual trees.
Carbon and biodiversity figures are calculated from the same sensor data the dashboard shows. When credits are issued, they get recorded against the registries (Verra VCS, Gold Standard) so the same credit cannot be sold by anyone else.
- Sequestration calculated from measured tree growth, not estimates
- Issuance into Verra VCS and Gold Standard
- Biodiversity Benefit Units aligned to the Kunming-Montreal framework
- Registry sync prevents the same credit being double-sold
No claim forms. No adjusters.
Drought, fire and tree-health thresholds are written into the policy at the start. When the sensors show a threshold has been crossed, the payout goes out automatically — to whoever the policy says it should go to.
- Triggers and payout amounts agreed up front
- Sensor data on the trees, not satellite estimates
- Underwriting being arranged with South African insurers
- Payouts by MoMo or EFT depending on the recipient
A board pack you can defend.
The four TNFD pillars — Governance, Strategy, Risk, Metrics — pull from live data instead of last year's estimates. Each figure has a citation trail back to the sensors and records that produced it.
- TNFD v2 disclosures across all four pillars
- Each metric links back to the sensor data it came from
- PDF for board packs, JSON for assurance teams
- Maps to JSE Climate Disclosure where required
The rules built into the way the system works.
Personal information is never stored in the clear. Each healer has a tenant-specific identifier derived from a salted hash. Consent is captured when they sign up, and the audit trail keeps a record of every change.
- POPIA Section 18 consent recorded per contributor
- Identifiers cannot be linked across corporate tenants
- Right-to-erasure available without losing the aggregate history
- Bioprospecting permits checked at the moment a harvest is declared
South African data, on South African servers.
Compute, storage and the models we use all run in our Johannesburg data centre. There is no failover region in another country, and no third party touches the data on its way through. Everything is governed by South African law.
- Single-region hosting in Johannesburg by choice
- Encrypted in transit and at rest
- Signing keys held in our self-hosted key management system
- Information Officer registered with the Information Regulator