Search the GenAI Directory

You have a photo of a struggling plant and want a first opinion on what is wrong. This is a small hosted web page where you upload a leaf image and get a suggested disease label, with the model also downloadable.

Water stress decisions on French farms still come down to judgement calls each week. This is the second version of a downloadable watering advice add on for a language model, trained partly on generated water stress examples.

French growers working out how many millimetres of water to apply have to combine soil type, crop stage and probe readings by hand. This is a downloadable add on for an existing language model that returns a watering amount in French.

When you are gathering farm data to train or test a model, every extra source helps. This is an agricultural dataset hosted on a model and data sharing hub, with little described beyond being available to download.

The same maize variety can do well in one field and poorly in another, and sorting out why takes many trials. This is a downloadable model that learns from sequence data to predict how maize responds to different growing conditions.

Farming questions often need an answer drawn from a pile of reference material rather than invented on the spot. This is open code for an assistant that searches a farming knowledge base, including linked relationships between facts, before it answers.

Growing without soil raises steady questions about nutrients, water and how to set things up. This is a preconfigured chat assistant, opened through a shared link, focused on hydroponic and soilless growing methods.

Brand managers wanting to know why sales moved usually file a request and wait for an analyst. This is a chat-style analyst that runs questions against natural products retail data and returns answers, charts and an explanation of what drove the change.

Training an image model on crop photos falls over quickly if you only have a few thousand pictures. This is a downloadable corpus of about two and a half million crop field images, split into parts, gathered for image generation work.

Following cattle, grain and milk prices in Brazil means checking many pages every morning. This is a server giving AI assistants lookup tools for those quotes plus weather, currency, export and harvest calendar figures.

When an assistant needs to answer a food question, it needs somewhere to look the facts up. This is an openly available plug-in interface that packages an open nutrition dataset so AI tools can query it directly.

You want soil properties for a field or region without writing code against a mapping service each time. This is a plug-in interface for AI assistants that lets them look up global soil data, with a hosted address you can point at.

This is a locally run MCP server template for querying USDA Foreign Agricultural Service Production, Supply and Distribution data from compatible assistants.

A grower or adviser with a quick crop question may not want to open a textbook. This is a small hosted chat demo for general farming questions, published together with its files so you can run your own copy.

Before trusting an AI system on a farming or food task you need a way to check it. Runs the ruTXTAgroBench agronomy task on lm-evaluation-harness with fixed prompts and packs results for leaderboard submission.

Before trusting an AI system on a farming or food task you need a way to check it. Ranks submitted models on sector benchmarks for agriculture and medicine using fixed run parameters and expert analysis.

Extension material sits in thousands of scattered fact sheets that assistants and search tools cannot use as they are. This is a programme and platform that organises that public extension content so institutional assistants can draw on it.

Working out which crop is growing in each Swiss field from satellite pictures is otherwise a survey job. This is a downloadable model, built on an existing earth observation model, that labels crops across Swiss farmland.

Buyers who need to sanity check commodity costs often juggle price sheets, forecasts and analyst notes. This is a hosted plug-in interface offering more than twenty commodity research tools, including price history, forecasts, scenario tests and cost breakdowns, to AI assistants.

Spray timing, irrigation and harvest calls all hinge on weather you can pull into your own tools. This is a commercial weather data interface covering past, current and forecast conditions worldwide, with farming among its listed uses.

Checking Korean food and drug safety records usually means visiting several government websites in turn. This is an installable add-on for an AI assistant that looks up those public Korean food safety data sources for you.

Field tasks get written down late, and dashboards need somebody technical to build. This is a hosted assistant reached mainly through WhatsApp that logs jobs by voice, builds dashboards from chat and links to farm and irrigation systems. The publisher describes it in the context of AI agents, AI Dashboard Builder, AI Workflows.

Dairy planning starts with knowing how many animals are actually being milked. This is a downloadable table of milk animal counts for African countries, arranged for direct use in models and scripts.

Mistral AI’s open-weight small multimodal model family for local and edge use. It includes 3B, 8B and 14B base, instruct and reasoning variants with image understanding and multilingual capabilities. It is a general foundation family rather than an agrifood-specific model.

Before trusting an AI system on a farming or food task you need a way to check it. Fruit detection, segmentation and counting in apple orchards using over 40,000 annotated instances and a separate test set.

Strategy questions about a category or a shopper group send you back into a pile of research reports. This is a hosted platform that answers those questions from a research publisher’s own library and links to the reports behind each answer.

Before trusting an AI system on a farming or food task you need a way to check it. Vision language models on agricultural expert consultation tasks covering identification, causal reasoning and multi-turn clarification over user submitted images.

Before trusting an AI system on a farming or food task you need a way to check it. Ranks models on single-turn and multi-turn agricultural consultation tasks using identification accuracy, reasoning quality and management recommendation metrics.

When you are comparing options for a farm advice assistant, general models belong in the comparison. This is a general purpose language model, offered through a programming interface, that has been evaluated on agricultural advisory questions.

Before trusting an AI system on a farming or food task you need a way to check it. Predicts soybean yield from genotype and weather information, with public and private submission leaderboards.

An open machine-learning compiler and deployment engine for running language models through a shared runtime across Python, JavaScript, iOS, Android and web applications. It exposes OpenAI-compatible interfaces while compiling models for local target hardware.

Watching dairy cattle closely usually means juggling cameras, collars and spreadsheets. This is a downloadable dairy cow monitoring dataset combining sensor readings and video with labels, test scripts and trained weights.

Getting software to recognise a dish and estimate what is in it needs a lot of food photos. This is a downloadable image collection of food, put together for recognition, sorting and nutrition related work.

Planning a breeding programme means guessing which crossing plan pays off before you commit years of field work. This is a simulator you run in the R language, letting you compare the genetic and cost outcomes of different programme designs.

When you need to know how a crop and its soil will behave season by season, guesswork is not enough. This is a downloadable simulation model that tracks plant growth alongside soil carbon, nitrogen and water movement.

Chaining together the steps behind a watering recommendation, from data to advice, is fiddly to wire up yourself. This is a ready made automation workflow in French that runs several cooperating AI steps over farming and water balance questions.

Handwritten field notes and scanned paperwork still need someone to type them up before anything can use them. This is a general language model that also reads images, so software can pull text and meaning from photos and documents.

You would like to ask an assistant what is in a packaged food instead of squinting at the label. This is a small plug-in interface connecting Claude to a community food products database so it can answer nutrition questions.

Before trusting an AI system on a farming or food task you need a way to check it. Predicts cucumber crop phenological growth stage among five classes from field images, ranked by macro F1 score.

Assessing a farm loan means reading many statements and documents before you can judge the risk. This is a hosted agent for agricultural credit review that lenders and traders can also call from their own systems.

Not knowing how a growing season might turn out makes planning and buying decisions harder. This is a hosted demonstration app that estimates crop risk for an area, with its files published so you can inspect or rerun it.

Crop models and advisory tools need daily weather and sunlight figures for places with no nearby station. This is a free public service serving global weather and solar records, including a section arranged for agricultural and climate uses.

You need code that can name the dish in a photo without training a model yourself. This is a downloadable image classifier, adapted from a general vision model and trained on a public collection of everyday dishes.

A farmer without a smartphone or nearby adviser may still need an immediate answer about pests, weather or crop management. The National Hotline for Agriculture is a toll-free AI voice service that gives personalized advice in English, Hausa and Yoruba.

Recognising plant disease from a leaf photo takes many labelled examples across crops and symptoms. This is a publicly hosted image dataset of diseased and healthy plants, usable for training, checking or reference.

Local breeds and husbandry practices in Nigeria rarely feature in general model testing. This is a set of questions on Nigerian indigenous livestock knowledge, grouped by topic, for checking what a language model knows.

Testing a herd record system needs realistic data when the real thing cannot be shared. This is a downloadable made up Nigerian livestock inventory listing animal type, breed, age and location for each row.

Proving where an animal has been is central to traceability, and real records are sensitive. This is a downloadable set of invented Nigerian livestock movement and traceability records for building and testing systems.

Pulling a retail measurement report means knowing the platform’s menus as well as the question you want answered. This is an assistant inside a retail measurement platform that answers questions in plain words, turns them into report queries and writes up the findings.

Blanket spraying a field uses more chemical than the weeds actually need. This is a product that fits cameras to machinery you already own so the sprayer only fires where it is needed.

Row crop growers have questions about products, seed choices, pests, equipment and now marketing, and few places to ask all of them. This is a web advisor built on farm agronomy writing, product labels and public weather and soil sources. The publisher describes it in the context of FBN AI Ag Advisor.

New staff in a flower shop ask the same questions and your procedures live in scattered documents. This is a paid online course showing you how to build a training assistant from your own notes and standard procedures.

Growers have questions at odd hours and records that never get typed up. This is a family of hosted Japanese farming assistants, one for round the clock agronomy questions and one that captures work, growth and pest records.

Choosing rams and planning matings means comparing breeding figures and checking relatedness across many animals. This is an open command line program and plug-in interface for AI assistants that searches sheep breeding records and compares, ranks and pairs animals.

You want one DNA model you can point at plant work as well as animal work. This is the third version of a downloadable genome model trained across twenty four animal and plant species, including a version that can generate sequence.

Comparing genome models needs a fixed set of tasks so results mean the same thing each time. This is the downloadable task collection shipped with a genome model, covering both animal and plant sequence problems.

Genomics work often starts with needing a general model of DNA sequence to build on. These are downloadable model weights and open code for a family of genome models you can run or adapt for your own predictions.

Field observations are often lost because typing is impractical while walking or operating equipment. Numanac is a voice-first farm management app that structures spoken logs, supports questions over the resulting records and states that full product functionality continues offline before syncing on reconnect.

People ask assistants how many calories or carbohydrates are in a meal, and there is no easy way to check the replies. This is a benchmark of meal descriptions checked by people, with nutrition figures to score model estimates against.

Estimating what is on a plate from a picture needs meals measured properly first. This is a downloadable collection of thousands of meals with depth images, video, ingredient lists, weights and nutrient values.

Reading a nutrition label and judging how healthy a product really is takes practice. This is a preconfigured chat assistant, opened through a shared link, that goes over labels and comments on food products.

Asking an assistant about calories or ingredients means it needs somewhere official to look them up. This is a small Python service that connects an assistant to United States government nutrition records, with meal planning prompts included.

You need lots of realistic video of physical situations to develop and test machines that move through the world. This is a family of downloadable models that generate and reason about video of physical scenes.

Teaching a field robot to steer or pick without wrecking real hardware first is the hard part. This is a code framework for training robot control policies inside a simulator, using trial and error or copied human demonstrations.

Testing machinery in a real orchard or field takes a season and a budget you may not have. This is simulation software where you build physically realistic 3D scenes, run virtual robots in them and generate practice images.

Deciding whether a field is heading for water stress means weighing several signals at once. These are downloadable model weights, packaged for local running, for a text model that talks through irrigation and water risk questions.

Questions about leasing livestock and the services around it need someone who knows that particular offer. This is a preconfigured chat assistant, opened by shared link, answering about one company’s livestock leasing and farm services.

Choosing between genome models is guesswork unless you can see them run the same tasks. This is a set of shared tests with downloadable data, code and a public scoreboard, including plant DNA and RNA tasks.

Some plant sequence work needs the shape a molecule folds into, not just the letters. This is a family of downloadable plant DNA and RNA models, in several sizes, trained to line up sequence with structure.

Collecting enough labelled barn footage to train cattle spotting software is slow and awkward around live animals. This is a set of 3D barn scenes with a pipeline for making synthetic livestock images, plus two ready trained detection models.

A mobile-app agronomy agent that interprets a specific field using OneSoil satellite, weather, crop-history and operations data to provide daily ranked priorities.

Studying how plant shape drives light capture and growth usually means stitching together many small scripts. This is a Python platform with linked components for modelling plant architecture and growth, so the pieces fit into one workflow.

Gene editing work in crops relies on a small set of natural editing proteins. This is a gene editing protein designed by a language model rather than found in nature, released with matching guide molecules and ordered as physical material.

Checking whether an ingredient or product has been recalled is a task nobody wants to do by hand. This is an openly licensed plug-in interface that lets AI assistants search regulator recall notices and reported adverse events, including food.

Looking up what is in a packaged product, across many brands and countries, is not something you can do by hand. This is an open database of food products, ingredients, nutrition figures and photos, available through a data interface or bulk download.

Training anything that reads food packaging needs many pictures of real products. This is a large downloadable image collection of packaged foods, mirrored in a public open data registry on a cloud provider.

Beyond reading packaged food records, you sometimes need to correct them or add a photo. This is an openly licensed plug-in interface giving AI assistants barcode lookup, product search, autocomplete, and the ability to edit entries and upload images.

You want an assistant that can pull up packaged food details by name or barcode. This is an open plug-in interface that makes a community run food product database available to AI tools without you writing the query code.

Getting the figures off a nutrition label into a database is tedious and easy to get wrong. These are downloadable model weights, with code, that pull nutrition values out of food packaging information.

Checking what is actually in a packaged food, including allergens, means reading labels one at a time. This is a downloadable database of food products with ingredient lists, allergen flags and nutrition figures, packaged for use in code.

Anyone pre training a DNA model needs a very large, cleaned sequence collection, and building one is a project in itself. This is a downloadable corpus of trillions of DNA letters from over one hundred thousand genomes, including plant material, split into sections.

Advice written for another country or another crop is not much help when you are standing in your own field. This is a free advisory assistant that searches extension guides, journals and grower organisation material filtered to your region and crop, and cites them.

Calls to crews, buyers and suppliers eat into a grower’s day and often just repeat the same information. This is a voice assistant wired into a specialty crop record system so those calls and follow ups can run on their own.

Looking up nutrition facts for foods one at a time slows any project down. This is a downloadable foods and nutrition dataset, also reachable through a data site interface, usable for training, grounding or testing.

Identifying young plants from seedling photos is a common research need, and labelled examples are limited. This is a public image dataset of seedlings with the species recorded for each plant, free to download and use.

What roots do underground shapes water and nutrient uptake, and digging them up to look is destructive. This is a simulation model of root architecture and resource capture that you can run instead of excavating.

Fruit trees need decisions on pruning, varieties and care at particular times of year. This is a preconfigured chat assistant, opened through a shared link, focused on orchards, fruit growing and horticulture.

Purchase orders arrive as attachments, scans and handwriting, each customer with a different layout. This is a hosted agent that reads them, checks the contents and turns them into order records, learning the layouts it sees repeatedly.

Keeping crop decisions consistent across every field is hard when growers and agronomists cannot visit each plot every day. This is Aydi’s AI agronomy platform, which uses plot records, photos and crop context to help users ask about disease, water, spray timing and field work. It has a free sign-up tier and paid plans.

Food distributors spend their days chasing orders, suppliers and unpaid invoices by phone and email. This is a hosted set of assistants that work together on order handling, buying, supplier messages and credit chasing.

Farm financial planning software holds your numbers but answering a question still means building a query. This is a built-in assistant in an Australian farm business planning tool that takes questions about your figures in everyday language.

Writing advertising copy for a farm product or service from a blank page is slow going. This is a set of seven prompts that walk a chat assistant through planning and drafting paid advertising campaigns for agricultural offers.

A farmer facing a pest, a fertiliser question or an odd weather week wants a starting point rather than a textbook. This is a chat assistant that gives first suggestions on planting, fertiliser amounts, pests, diseases and weather.

A lightweight open vision-language model family combining a SigLIP vision encoder with Gemma language models. It supports image and text input for captioning, visual questions, object detection and text reading, with 3B, 10B and 28B sizes. Pretrained checkpoints require task-specific tuning; mixed-task checkpoints are intended mainly for exploration and research.

You want meal images with nutrition details already attached rather than measuring plates yourself. This is a downloadable copy of a meal and nutrition dataset, shared with code for training, grounding or testing models.

Arranging partnerships or working with rural voices online means drafting a lot of near-identical messages. This is a set of six prompts for a chat assistant covering partnership and influencer tasks in agriculture.

You want to estimate crop growth and yield for a site and season without installing old command line tools. This is a crop simulation library written entirely in Python, carrying several established grassland and arable growth models.

Designing a greener crop protection molecule means moving through target choice, ideas, tweaks and safety checks. This is a web platform that generates candidate molecules from the shape of a target pocket and predicts binding strength and effects on the environment.

Researchers analyzing plant genomes can use a hosted prediction page or run the accompanying code and models. PDLLMs is a collection of plant DNA language models with online prediction, training and inference resources.

Coming up with new crop protection molecules to try is slow chemistry work. This is a web service where you upload a protein target and get back generated candidate molecules along with predicted binding, properties and how hard they look to make.