What are AI assistants and how to use them
Ever since OpenAI released their ChatGPT in November 2022, the number of AI assistants have sky rocketed. In a series of three Agtech Letters we give an introduction to important opportunities, risks, and technology related to AI assistants. The purpose is to clear some of the fog and increase your curiosity.
AI assistants are good at different things
In the next Agtech Letter we will describe the technology used in AI assistants and how the different parts are connected. In this Agtech Letter, its sufficient to describe the AI assistant as an AI model that the user can ask questions or instruct to do tasks using so called prompts.
Most of the tech giants are developing own AI models that can be used in their AI assistants, or connect to other services to create third-party AI assistants powered by their AI models. Many of the AI assistants from the tech giants are generic, meaning that they are trained to answer questions and perform tasks related to a broad range of topics. However, there are also companies that develop AI assistants that are tailored to specific tasks or topics, for example an AI assistant that is specifically trained for agriculture.
Choosing the right AI assistant suitable for the task at hand is important to get good and reliable support from it!
Some examples of AI assistants
Today there are several types of AI assistants on the market, both the free versions that have gained lots of popularity but also more specific AI assistants than have been trained to solve specific tasks. We list some of the currently available AI assistants that are interesting and known to us.
Examples of generic AI assistants
- Gemini – Google’s AI assistant. Based on their AI model Gemini 1.5.
- ChatGPT – OpenAI’s AI assistant. Based on their AI model GPT-4.
- Claude – Anthropic’s AI assistant. Based on their model Claude 3.
- Copilot – Microsoft’s AI assistant. Based on OpenAI’s model GPT-4.
The different AI assistants listed above are all good at different things and the development is proceeding at a very high speed and hence the performance is increasing rapidly as well. All of the listed AI assistants are powerful tools that are also available in free versions, but to get access to the best AI models the user usually have to pay some kind of license fee.
Examples of AI assistants for agriculture
- Norm – An AI advisor developed by Farmer Business Network.
- Taranis Ag Assistant – An AI-driven agronomist advisor from Taranis.
- Headstorm AgPilot – An AI-driven agronomist advisor from Headstorm.
These represent a selection of some AI assistants that has been specifically developed for agricultural purposes and might be worthwhile to try out. At least to give you an idea of the differences between a generic, yet powerful, AI assistant from the big tech companies and a specific AI assistant targeted to agriculture.
How can you use and benefit from AI assistants?
AI assistants can help out with lots of things, but to give a simple yet illustrative example, let’s consider to use an AI assistant for helping out with financial calculations to determine whether the crop production business is profitable or not. This will allow us to highlight important steps in the interaction with the AI assistant than can be good to know about.
In this example, the free version of OpenAI’s AI assistant ChatGPT is used. Feel free to try the example yourself and see if you get a similar result.
Caution: You should never provide the AI assistant with sensitive information such as personal data in the free versions.
In the example, I start by giving ChatGPT some background information to help it answer in a relevant way and then I ask it to help me with the steps for performing an economic analysis of my crop production business. Writing these types of instructions and questions to the AI assistant is called prompting and is a central concept for AI assistants. It is important to prompt in a good way to get relevant and high-quality answers back, and this process is sometimes denoted prompt engineering.
After I’ve got the answer, I ask ChatGPT to dive into ways of improving the profitability without increasing land or employees and to maintain a moderate risk level in the business. This type of conversation is a good way to interact with the AI assistants and to improve the quality and precision of the answers.
The chat is provided below.
Example: ChatGPT helps out with profitability analysis
Me:
I’m a crop farmer in the UK and wants to analyze the revenues and profits from my fields. The purpose is to determine whether my crop farming business is profitable or not. Can you please provide the major steps I need to do in such an analysis and where I might need help from an agronomist.
ChatGPT:
Analyzing the profitability of your crop farming business involves several key steps. Here’s a structured approach:
1. Data Collection
- Yield Data: Record the harvested quantity per field (tons/ha).
- Crop Prices: Gather market prices or contract prices for each crop.
- Input Costs:
- Seeds
- Fertilizers and pesticides
- Fuel and machinery costs
- Labor costs (hired and personal)
- Fixed Costs:
- Land rent/mortgage
- Machinery depreciation
- Insurance, administration, and overheads
- Other Costs:
- Irrigation costs
- Storage and transport fees
- Interest on loans (if any)
2. Revenue Calculation
- Multiply the yield per hectare by the crop selling price.
- Consider subsidies, grants, or incentives applicable in the UK.
3. Cost Analysis
- Sum up variable costs per hectare.
- Distribute fixed costs across all fields to understand per-hectare expenses.
4. Profitability Calculation
- Gross Margin per Crop = Revenue – Variable Costs
- Net Profit per Field = Revenue – (Variable Costs + Fixed Costs)
5. Breakeven Analysis
- Determine the minimum yield or price needed to cover costs.
- Evaluate which crops are most profitable under different conditions.
6. Field Comparison
- Identify high and low-performing fields.
- Assess whether certain fields should be used for different crops or alternative uses.
7. Long-Term Trends & Strategy
- Compare multi-year data to assess trends in profitability.
- Adjust crop rotations, input levels, or market strategies accordingly.
Where an Agronomist Can Help
- Yield Improvement: Identifying factors affecting productivity (soil quality, disease, pest management).
- Cost Optimization: Suggesting efficient input use or alternative cropping strategies.
- Sustainability: Recommending practices that improve long-term profitability.
- Precision Farming Insights: Using satellite data, soil analysis, or AI-driven tools for better decision-making.
Would you like help with structuring a spreadsheet or software tool for this analysis?
Me:
Thanks, I’m a bit curious on how to improve my profitability without scaling my operations in terms of land and employees. Can you point me to the top three actions to take to improve profitability while maintaining a moderate risk level.
ChatGPT:
Here are three key actions to improve profitability without expanding land or labor, while keeping risk at a moderate level:
1. Optimize Input Efficiency (Cost Reduction)
- Precision Farming: Use data-driven insights (e.g., soil testing, satellite imagery, variable rate application) to optimize fertilizer, pesticide, and irrigation use.
- Targeted Input Application: Apply fertilizers and pesticides only where needed to reduce waste and lower costs.
- Bulk Purchasing & Alternative Suppliers: Secure better prices by buying in bulk or sourcing more cost-effective suppliers.
2. Improve Crop Selection & Rotation (Revenue Maximization)
- Higher-Value Crops: Introduce more profitable or specialty crops that fit your soil and market demand (e.g., niche grains, cover crops for soil health).
- Strategic Rotation: Implement a rotation plan that boosts soil fertility, reduces pest pressure, and maximizes long-term yields.
- Direct Sales & Contracts: Consider selling directly to consumers, food processors, or securing forward contracts to lock in favorable prices.
3. Leverage Technology & Automation (Efficiency Gains)
- Yield Monitoring & Data Analysis: Use farm management software or AI tools to analyze past performance and make data-backed decisions.
- Smart Machinery & Drones: Implement automation (e.g., GPS-guided equipment, drone scouting) to reduce fuel, labor, and maintenance costs.
- On-Farm Processing & Value Addition: Explore small-scale processing (e.g., milling grains, packaging) to capture more value from your produce.
Would you like help identifying specific technologies or financial tools that fit your operation?
Some conclusions from the example
In the example, ChatGPT helps us understand what is driving profitability, what the building blocks of the profitability analysis is based on, how to compute the profitability analysis, and a few tips to improve the analysis. Keep in mind that this is from a generic, free, AI assistant that is not specifically trained on agriculture.
On the question on how to improve the profitability, ChatGPT lists a few suggestions to increase revenues and or decrease costs.
As can be noted in the responses from the ChatGPT, all information is structured in a certian ways with a header of information, followed by bullet-point lists of information, and wrapped up by a small summary and questions on next steps. Furthermore, all information is described quite general and without details. If more details are required, then follow-up questions and or corrections in new prompts can be provided. For example, you can ask ChatGPT to elaborate more on how Smart Machinery & Drones can help you improve profitability. Is the cost of the equipment worth it? etc.
How reliable the answer from the AI assistant is depends on the circumstances such as the prompt, the choice of AI assistant, the skill of the person using the AI assistant and much more. In the third part of this series on AI assistants we will show some cases where the AI assistant answers incorrectly together with some key information that you need to understand when using AI assistants in your business.
Agtechers' Actions
Trying a few different AI assistants doesn’t cost you more than the time invested, and using them to your benefit could become very valuable to you. Hence, our recommendation is that you spend some time to get to learn using AI assistants, but keep in mind to not share sensitive data with the free versions.
In the next Agtech Letter we will describe the technology that is powering AI assistants.
