Agent Crop

Agent Crop is a cannabis strain identification tool. Agent Crop analyses uploaded crop and plant photos with computer vision to detect disease and identify species, then recommends treatments for what it finds. Runs in the browser, accepts PNG and JPG files up to 50 MB total, and clears images automatically after each detection.

Agent Crop is a web tool built by a Microsoft for Startups Founders Hub member with a single, narrow focus: turning a photograph of a plant into useful diagnostic information. The platform runs computer vision models that look at uploaded images of crops and plants and return two kinds of output. The first is disease detection, which flags likely conditions found on the plant. The second is plant identification, which names the species or crop in the frame. For any issue the model detects, the tool also surfaces recommended treatments, so the result is not just a label but a starting point for action. The workflow is deliberately plain. A visitor opens the site, chooses between the disease detection and plant identification modes, and uploads images through the browser. Files must be PNG, JPG or JPEG, and the total upload per run is capped at 50 MB. There is no account gate described on the homepage and no software to install, which keeps the barrier low for someone standing in a field with a phone photo. A Cloudflare human check runs before processing, which suggests the team is guarding the endpoint against automated abuse rather than inviting bulk programmatic access. Privacy is handled in a way that suits agricultural users. The homepage states that uploaded images are not stored and are cleared automatically once detection finishes, so growers do not have to worry about a permanent archive of their field photos sitting on a third party server. That is a meaningful detail for anyone photographing proprietary cultivars or uncovered pest problems. This is a practical triage aid rather than a replacement for an agronomist or a lab test. The audience is growers, smallholder farmers, home gardeners, agronomists doing a first pass in the field, and agricultural students learning to recognise symptoms. Among general purpose image recognition tools, Agent Crop stands out by narrowing to one vertical and pairing the diagnosis with treatment guidance instead of leaving the user to search for what to do next.