Search the GenAI Directory

Fish farming knowledge in Russian is not something general model tests cover. This is a Russian multiple-choice question set on commercial fish farming, presented online for checking what language models know about the trade.

Kitchen staff need reminders on safe food handling and a route towards certification. This is a preconfigured chat assistant, opened through a shared link, that coaches on handling practice and what certification involves.

Assistants sometimes give advice that is fine in general but unsafe in a particular kitchen or for a particular pet. This is a portion of a safety test collection covering food and pet food scenarios across many reworded questions.

Cold outreach to growers, dealers or co-ops means writing much the same emails and call notes again and again. This is a set of seven ready-made prompts that get a chat assistant drafting agricultural sales outreach.

Scoring how much of a leaf a disease has taken over is slow and differs between people. These are downloadable model weights that outline disease spots on leaf photos, tuned from a general image outlining model.

Spotting a leaf disease early is hard when you only have a phone photo and no specialist nearby. These are downloadable model weights, with code, that put pictures of diseased plants into named disease categories.

Training a farming assistant on general web text leaves gaps in agricultural language. This is a collection of farming text prepared for teaching a language model the vocabulary and topics of the sector.

You want to prototype a question answering tool over agricultural statistics but do not want to start from a blank file. This is sample code showing how to load United Nations farm statistics and have a language model search them before answering.

Damaged or odd looking bunches are rare in your photo archive, so a defect spotter has little to learn from. This is a code pipeline that creates artificial images of flawed table grapes to pad out training sets.

Seed and breeding questions are specialised enough that general chat assistants often stumble. This is a downloadable question set, checked by specialists, for measuring how language models handle seed science and plant breeding work.

A public MCP server over US row-crop seed catalog and yield-trial data. It exposes variety identity, multi-vendor trial results and a committed retrieval-evaluation harness to the Drawbar advisor.

A grower asks about cedar apple rust and you want an answer tied to real source material rather than guesswork. This is a code repository for a question answering system that searches a curated disease knowledge base before it replies.

Readings from field weather stations are more useful when they feed your own tools rather than one app. This is a data interface for on farm weather stations, giving device readings plus summarised and mapped weather data.

Pulling satellite pictures for a field means dealing with large archives and awkward file formats. This is a set of standard mapping web services for requesting Earth observation imagery from code, including the public European satellite programmes.

Nutrition and ingredient wording buried in regulatory text has to be picked apart before it is usable. These are downloadable model weights, with code, that tag nutrition related terms inside food and labelling text.

Keeping up with food rules and safety alerts in every market you sell into is a full-time reading job. This is a hosted platform for food risk and regulatory work, where you ask questions in plain language and watch recalls, border rejections, inspections, hazards and forthcoming rules mapped to your own products, markets and suppliers. The publisher configures it around your operation rather than handing you a login, and it offers programming access so the same signals can feed a product lifecycle, resource planning or supplier system you already run. Published previously as Digicomply.

Before trusting an AI system on a farming or food task you need a way to check it. Predicting sorghum phenotypes and genetic markers from RGB and laser scanner imagery, with loaders for defined benchmark tasks.

Before trusting an AI system on a farming or food task you need a way to check it. Predicts a complete 3D fruit mesh from partial RGB-D observations of sweet peppers, scored by F-score and Chamfer distance.

Policy and research work on African livestock needs species mix figures in a form you can load straight into code. This is a tidied table of each species share of total livestock across African countries, drawn from United Nations food statistics.

Shennong V4.0 is a hosted agricultural foundation-model assistant covering crop and livestock questions, planting guidance, pest and disease image analysis, yield, soil, weather and agricultural knowledge. The publisher invites organisations to apply for deeper trial access.

Nutrition questions in Chinese are a blind spot in most model testing. This is a nutrition slice of a Chinese multiple-choice exam collection, packaged for download so you can score models on those questions.

A multilingual vision-language encoder family for zero-shot image classification, image-text retrieval, localization and dense visual features. The downloadable checkpoints can also serve as the vision encoder inside a VLM. It is general-purpose and not agrifood-tuned.

Apps that need product details from a barcode usually end up stitching together several sources. This is a food data service with endpoints for barcode lookup, search and product details, including nutrition and allergen information, plus a plug-in interface for AI assistants.

Chinese farming questions span horticulture, plant protection, breeding and rural economics, and general models have read little of that writing. These are downloadable language models in two sizes trained on billions of words of Chinese agricultural literature, released with code.

Deciding when and how much to irrigate means pulling weather, soil and crop water figures together each week. This is an importable automation workflow that gathers those inputs and produces an irrigation schedule for a farm.

Lending and insurance questions from small farmers need answers you can sanity check before going live. This is a small held-back question set shipped alongside a fine-tuned add-on for smallholder farm finance, for testing its replies.

Livestock advice is needed in sheds and yards where an internet connection cannot be relied on. These are downloadable adapter weights and a packaged file for running a livestock focused language model on your own machine.

A small open vision-language model family for image, multi-image and video understanding in 256M, 500M and 2.2B sizes. Hugging Face reported an internal local iPhone demonstration with the 500M model. The downloadable model is current, but this listing does not represent a publicly obtainable iPhone application.

Before trusting an AI system on a farming or food task you need a way to check it. Computer vision models predicting foods, ingredients and nutrients from photographs of meals, snacks and beverages paired with dietary records.

Soil science answers are scattered across decades of published papers. These are downloadable model weights, with code, for a language model that received extra training on soil science literature.

Before trusting an AI system on a farming or food task you need a way to check it. Requires soil organic carbon models to pass bias, RMSE and R squared tests on independent validation data, reviewed by validation bodies.

Soil questions cross chemistry, physics and biology, and the answers are spread across textbooks. This is a preconfigured chat assistant, opened through a shared link, for discussing soil science and the tools used to study it.

Drone and aircraft images of trial plots pile up faster than anyone can turn them into numbers. This is a programming interface for uploading field imagery and getting back stitched maps, elevation, plant counts and per-zone statistics.

Before trusting an AI system on a farming or food task you need a way to check it. Cultivar identification from sorghum plant images, with a fixed test set and reported top-1 classification accuracy baselines.

Before trusting an AI system on a farming or food task you need a way to check it. Predicts end-of-season sorghum biomass from visual data in plant images.

Before trusting an AI system on a farming or food task you need a way to check it. Ranks models predicting sorghum genotype from plant imagery, reporting top image accuracy and top cultivar accuracy per marker.

Soybean work often needs example data before you can test an idea. This is a soybean dataset listed in an open catalogue, downloadable for training a model, giving it reference material or scoring its output.

Tracing an allergen from an ingredient through a formula to the finished pack means opening record after record. This is a chat assistant inside a specification system that answers plain questions about ingredients, formulas, packaging and certificates and links to the records.

Deciding which flavour or ingredient to build a product around is mostly guesswork without data. This is a hosted application for food and drink teams that reports on trends and ingredients and sketches product concepts from them.

Insight teams pay for market reports and then struggle to find the one chart that answers today’s question. This is a hosted research assistant that sits over a large library of food and drink reports and answers questions and summarises across them.

A customer asks for a recognised food safety certification and you have to work out what that involves. This is a preconfigured chat assistant, opened by link, guiding food businesses through one such certification and keeping it.

Herd health records from sub-Saharan Africa are hard to obtain and share, which stalls model building. This is an artificially generated collection of livestock health and output records for that region, meant for training and experiments.

Farm loan files arrive as stacks of borrower and collateral documents that credit staff read line by line. This is a hosted set of assistants that pull out the details, check them against lending policy, draft memos and route files for decision.

Grain contracts, tenders and shipping paperwork all need reading and reconciling against your positions. This is a hosted set of assistants that lift the key terms out of those documents and take over some of the routine logistics admin.

Launching batches of crop model runs and reading the output files by hand gets tedious fast. This is an R package that drives an established crop model from your scripts and turns its output into tidy tables.

Supplier teams working with a grocery chain’s shopper data have to learn a reporting tool before asking anything. This is a chat feature inside that retail data platform that answers questions, shows where the numbers came from and suggests next steps.

Input companies field farmer questions, dealer messages, orders and paperwork with the same small team. This is a hosted assistant, trained ahead of time on farming material, that answers crop questions and handles ordering and rule checking steps.

Building voice tools for farmers means finding recordings and transcripts in the right language and subject area. This is a speech dataset about grain, shared for training or testing systems that turn spoken words into text.

Farming seaweed or fish at sea means decisions about systems, permits and the surrounding environment. This is a preconfigured chat assistant, opened through a shared link, covering marine crop growing, fish farming systems and environmental care.

Before trusting an AI system on a farming or food task you need a way to check it. Crop type mapping, crop yield prediction, field delineation and cropland segmentation tasks with PyTorch dataloaders and baseline models.

Regulators and quality teams need to point to the exact rule behind an answer, not a vague summary. This is a hosted assistant that reads your official documents and replies with references back to the source text. The publisher describes it in the context of specialised AI assistant for unstructured data.

Watershed and groundwater questions normally mean installing modelling software and finding the right dataset. This is a plug-in interface for AI assistants that opens up farm watershed models and groundwater records, with several tools available on a free account.

Shoppers increasingly ask chat assistants about products, and your product details may be scattered or out of date. This is a paid service that tidies, enriches and watches your product information, keeping a source and approval record on each detail. The publisher describes it in the context of generative engine optimization for product content.

When shoppers ask a chat assistant about ingredients or allergens, your product details need to be there and correct. This is a service that pushes structured product content, including nutrition and allergen fields, into ChatGPT and shapes it for other assistants.

Hand labelling thousands of photos so a model can spot objects is dull, slow work. This is an end to end pipeline built on a game engine that renders images and their labels together, ready for training.

Image sets for teaching software to outline plants in a field are small, and tracing every leaf by hand is slow. This is a code pipeline that invents plant pictures together with their outlines, in colour and near infrared.

Hand labelling tomato plant photos is tedious, so people try training on computer generated images instead. This is a downloadable set of synthetic tomato images for teaching models to outline plant parts, released with a study on adapting an image model.

Rural credit rules are confusing and getting an explanation often means a trip to the bank. This is a WhatsApp assistant, built on a general chat model, that answers farm credit questions for producers. The publisher describes it in the context of rural-credit WhatsApp assistant.

Food and drink teams need a steady flow of product and menu ideas, plus the write-ups to present them. This is a hosted assistant for the food and drink trade that drafts product, dish and menu suggestions along with reports and slide decks.

Food and drink teams want answers about what people are eating and why, without commissioning a study each time. This is a hosted platform of task specific agents built on the publisher’s own food and beverage data, covering trends, marketing, sales, product innovation, menus, recipes, surveys, concept testing and launch tracking, with a builder for making your own and connections into systems you already run.

Before trusting an AI system on a farming or food task you need a way to check it. Classifies crop condition as diseased, pests, stressed or healthy by F1, plus a fairness check across crop districts.

Crop input labels and product details differ by supplier, so agronomy apps end up doing manual cleanup. This is a programming interface that serves crop input product and label information in one consistent shape.

Breeding and crop research moves slowly when field measurements sit in separate places. This is an open reference collection of crop genetic data, detailed plant measurements and sensor imagery from field trials, free to download and reuse.

An AI-native, sub-acre agronomic decision platform that combines field data with the publisher’s deterministic agronomy framework to analyze field conditions and product performance.

You want to run a text generating model on your own machine, for example behind a small farm data service. This is a packaged set of downloadable model weights in a format built for local running, along with an example web service.

Farm decisions on yield and input spending are often made without the data being pulled together. This is an open code project that combines farm data with language models to produce forecasts and suggestions for developers to build on.

Deciding what to plant in an Indian field involves weighing many local factors at once. This is a preview language model that works through crop choice questions step by step, published as downloadable weights with code.

Specialty growers rarely have a finance chief watching cash, costs and market moves week to week. This is a hosted tool that keeps an eye on those numbers, flags risks and points out borrowing options for the farm.

Something is chewing the leaves and you want guidance you can trust rather than a search result. This is a public chat assistant that answers plant and pest questions from university extension and plant health reference material.

Telling tobacco plants from the weeds around them is the basis for targeted spraying and hoeing. This is a labelled image dataset of tobacco fields with weeds marked, listed in an open catalogue for download.

Running a chat assistant means paying for servers and wrangling model files unless someone else hosts it. This is a hosted API, used with an added key, that provides the language model inference behind the AgriGenius chatbot.

You want a chat or voice assistant for your agricultural business but no developers to write one. This is a hosted platform for building conversational and voice agents without code, with pages aimed at farming uses.

Predicting yields from satellite data is easier to work on when the data is already prepared and split. This is a crop yield forecasting dataset wired into a mapping focused machine learning library, ready to load in code.

Supplier specifications, allergen statements and lab certificates arrive as free form documents that buyers and quality staff read one at a time. This is an added layer inside a supplier management platform that turns those files into fields you can search and reuse.

Showing where a food ingredient came from, with paperwork to back it up, gets awkward when records sit with different suppliers. This is a hosted application that gathers traceability documents and builds a chain of custody record you can show.

Picking which crop lines to advance is easier with a trait prediction model, but building one takes time. This is a published code repository with a trait prediction model for crops and the scripts needed to train it on your data.

Moving stock by road brings rules on journey times, space and paperwork that vary from place to place. This is a preconfigured chat assistant, opened through a shared link, that explains Australian livestock transport requirements.

Finding overseas buyers for food and farm products means cold outreach in other languages and then a lot of paperwork. This is a hosted platform that searches trade records for likely buyers, runs the messaging, and drafts contracts and shipping documents.

Buyers and traders in food commodities need a read on prices, supply and trade policy without writing the report themselves. This is a set of market intelligence services, reachable by programming interface, that draft written reports and purchasing suggestions from trade data.

Machine records, field boundaries and application maps often sit locked inside equipment platforms when you need them in your own software. This is a set of connection points that let your code read and write that operational farm data.

Designing gene editing parts for plants means guessing which small sequences the plant will handle correctly. These are downloadable models trained on plant transfer RNA sequences that find those sequences and predict how they get processed.

A label review means checking claims, required wording and allergen statements against the current rules, and the rules move. This is a hosted review platform that looks up the regulation text as it checks and marks each finding on the artwork.

Specification sheets and lab certificates pile up, and allergen or labelling problems can hide in them. This is an added layer across two food compliance products that turns supplier documents into structured records and raises flags on the details it finds.

Food and beverage R&D teams make product, formulation, cost and compliance decisions across disconnected systems and old project files. Turing is a commercial, domain-trained AI and agent platform that connects product strategy, formulation optimization, cost trade-offs and institutional R&D knowledge in one hosted workspace.

Farmers and advisors asking questions in Turkish get little help from models trained mostly on English. This is a set of Turkish farming questions with answers, formatted for fine-tuning an open language model.

You need farm answers in Turkish covering local crops, livestock, subsidy rules and regulations. This is a downloadable language model tuned on Turkish farming texts and packed into a single small file, so you can run it on your own machine.

Recognising dishes in a photo, and finding where they sit on the plate, needs marked-up examples. This is a food photograph dataset covering 256 dish types with boxes drawn around each item.

UlangiziAI is a deployed agricultural advisory chatbot for smallholder farmers in Malawi. It delivers locally grounded answers through WhatsApp and SMS, including voice, text and image interactions; the cited page documents the programme rather than offering a public self-serve app.

Studying what drives corn and soybean yields over decades takes weather, soil and practice records in one table. This is a downloadable dataset of county yields for hundreds of Corn Belt counties since 1984 with weekly weather, soil and management fields.

Nutrient figures for foods get retyped from web pages into spreadsheets more often than anyone would like. This is a public data interface and bulk download of food composition records, also reachable by AI assistants through a plug-in interface.

Looking up what is actually in a food, nutrient by nutrient, means searching a government database yourself. This is a plug-in interface for AI assistants that connects them to the United States food composition records so they can fetch those numbers.

Knowing what produce and livestock are selling for, and what is shipping where, shapes selling decisions. This is a data interface to United States government market reports covering prices, shipments and movement of farm commodities.

Knowing which crop grew in which field across the United States is the starting point for a lot of land and yield analysis. This is a yearly map layer of crop cover for the mainland United States, available through a cloud imagery platform.

United States farm survey and census figures are useful, but the query system is fussy. This is an open plug-in interface that lets AI assistants ask for those statistics in plain language instead.

Working out acreage, production or price history for a crop and county means digging through survey publications. This is a data interface to United States government farm statistics that your code or spreadsheet can query directly.

You want a desktop assistant to fetch and compare real nutrition figures rather than recall them. This is a small server that opens a public food and nutrient database to AI assistants through a plug-in interface, with search and compare actions.

Separating crops from weeds by their light signature needs close-range spectral imagery that few groups can capture. This is a large collection of close-up multi-wavelength images of crops and weeds, shipped as numeric arrays.

Farmers hear that chat assistants save time but do not know what to actually type. This is a farmer-facing web page from a dairy body with worked example prompts and simple rules for improving the questions you ask.

Searching soil reports and papers is harder when they are written in different languages. These are downloadable model weights, with code, that turn soil related sentences into numbers so similar text can be matched across languages.

Old grain variety records arrive in inconsistent files that nobody can query as a whole. This is a data platform that loads them into one local database and exposes them through a plug-in interface for AI assistants and a breeding data endpoint.