We ran 5 real farm tasks through AI — here's where it saved time and where it flopped
Everyone's telling you AI will save you time on the farm. Grand. But which tasks? And does it actually work, or does it spit out nonsense that takes longer to fix than doing it yourself?
We sat down and ran five genuine Irish farm tasks through ChatGPT, Claude, and Gemini. Same prompts, same information, scored on the same criteria: accuracy, usefulness, and whether you'd actually use the output. Here are the honest results.
Task 1: Draft a TB test preparation checklist
The job: Create a checklist for preparing an 80-cow suckler herd for the annual TB test. Include facilities prep, paperwork, and animal handling.
ChatGPT: Produced a solid, well-organised checklist. Mentioned DAFM requirements correctly. Included facilities check, AIM system updates, and suggested sorting animals the evening before. Minor issue: referenced some UK-specific terminology ("CPH number" instead of herd number). Score: 7/10
Claude: More detailed on the paperwork side — flagged that you need your herd register up to date on AIM, mentioned checking your crush and handling facilities for safety compliance. Didn't include UK terms. Slightly long. Score: 8/10
Gemini: Shorter but accurate. Included a useful point about checking your water supply at the crush area that neither of the others mentioned. Missed the AIM system reference. Score: 7/10
Verdict: All three were useful. You'd spend 5 minutes tidying up the output versus 30 minutes writing it from scratch. Claude edged it for Irish-specific accuracy.
Task 2: Explain my ICBF EBI report in plain English
The job: We pasted in a real (anonymised) ICBF EBI report and asked each tool to explain it in plain English. What do the numbers mean? What should I focus on for breeding decisions?
ChatGPT: Good plain-English explanation. Correctly identified that fertility sub-index was the most economically important. Got the euro values per sub-index roughly right. Score: 7/10
Claude: Excellent. Broke down each sub-index, explained what a negative value in calving difficulty meant, and suggested which bulls to focus on based on the report's weak points. Flagged that EBI values change with base updates. Score: 9/10
Gemini: Decent but surface-level. Explained what EBI is but didn't dig into the specific numbers as deeply. Would be fine for a first-timer, less useful for someone who wants to act on the data. Score: 6/10
Verdict: Clear win for Claude on this one. If you're staring at an EBI report wondering what to do with it, paste it into Claude.
Task 3: Write a letter to the bank about a farm loan
The job: Draft a letter to the bank requesting a meeting about a €50,000 loan for a new slatted shed. Include farm details, purpose, and repayment ability.
ChatGPT: Professional, well-structured letter. Included sections for farm background, project details, and financial position. Slightly generic — could be any country. Score: 7/10
Claude: Similar quality. Added a reference to TAMS III grant eligibility which was a smart inclusion — banks like seeing that a grant covers 40-60% of the cost. Score: 8/10
Gemini: Also good. Slightly shorter. Mentioned attaching a cashflow forecast, which is practical advice. Score: 7/10
Verdict: All three produced usable letters. Claude's mention of TAMS was the most farm-business-aware addition. Any of them would save you an evening of staring at a blank page.
Task 4: Calculate feed costs for housing 80 suckler cows
The job: Estimate the total feed cost for housing 80 suckler cows for 120 days, assuming silage and meal supplementation at current Irish prices.
ChatGPT: Gave a reasonable estimate. Used silage costs of approximately €30-35/tonne and meal at €350/tonne. Came out at roughly €18,000–22,000 total. The assumptions were stated clearly. Score: 7/10
Claude: Similar calculation but broke it down per cow per day, which is more useful for management decisions. Also noted that silage costs vary hugely depending on whether you're making your own or buying in. Score: 8/10
Gemini: Got the rough ballpark right but used some prices that seemed UK-influenced (silage slightly cheap for Irish bought-in prices). Score: 6/10
Verdict: Useful from all three as a starting framework. But you must check the prices against your own actual costs — AI tools don't know what your contractor charges or what meal is at your local co-op this week.
Task 5: Summarise a Teagasc advisory bulletin
The job: We pasted a 2,000-word Teagasc bulletin on spring grazing management and asked for a 200-word summary with action points.
ChatGPT: Clean summary, hit all the key points. Three clear action items at the end. Score: 8/10
Claude: Slightly more detailed summary. Four action items, each with a timeframe. Added context about why early turnout matters financially. Score: 8/10
Gemini: Good summary but missed one key recommendation about target pre-grazing covers. Three action items. Score: 7/10
Verdict: Tie between ChatGPT and Claude. All three turn a long document into something you'd actually read at the kitchen table in 2 minutes.
Overall findings
| Task | Best tool | Time saved | |------|-----------|------------| | TB test checklist | Claude | 25 minutes | | EBI report explanation | Claude | 20 minutes | | Bank letter | Claude (just) | 45 minutes | | Feed cost calculation | Claude | 15 minutes | | Teagasc summary | Tie | 10 minutes |
Claude was most consistently useful for Irish farming tasks. ChatGPT was close behind and sometimes matched. Gemini was decent but less tuned to Irish specifics.
The honest take: AI saves genuine time on writing, summarising, and calculating. It does not save time on anything that requires seeing your farm, handling your animals, or knowing your specific soil. Use it for desk work. Don't trust it for fieldwork decisions without checking.
Sources
Was this useful?