Octara Flora
Research previewField biology that reports what it can prove.
Point a camera at a plant or a mushroom. Flora tells you what the evidence supports, how confident it is, what it could be confused with, and what is known about it. For wild fungi it never clears a specimen as safe to eat.
Flora does not guess.
Amanita muscaria
identified by FungiTastic + PlantCLEF + vision
How a scan resolves
Evidence first, then a name
No single model is the authority. Heterogeneous confidences are calibrated and fused, and the result backs off to the most specific rank the evidence justifies.
- 01
Capture
Point the camera; organ tags, focus, light and GPS are captured.
- 02
Identify
Multiple specialist models vote, each with calibrated confidence.
- 03
Reconcile
Evidence is fused; the result backs off to the rank it can justify.
- 04
Ground
Real occurrence data near the observation informs rarity.
- 05
Report
A field codex of taxonomy, ecology, pharmacology and hazards.
Multi-model identification
Several specialists, calibrated and fused
Confidence governance
It names only the rank it can justify
Rather than force a species, Flora reports the most specific rank the evidence supports and stops there. It never invents a species to fill the gap.
Fail-closed safety
A photo can never clear a wild mushroom
Edibility comes from curated mycological references, not from the model. A species absent from every list is unknown, never edible. Flora never emits a safe-to-eat verdict for a wild mushroom, even for documented-edible species, because toxic look-alikes exist.
If the identity is not resolved to species, no consumption verdict is offered at all.
Field codex
An evidence-grounded dossier, every claim sourced
Flora is an accuracy-first research companion. It is currently a self-hosted research preview: the identification models run on the researcher's own machine.