This small business provided villa maintenance. It had two non-technical managers but many labourers. The main headache: 150-odd site reports emailed to clients at every month’s end. Details sat in a spreadsheet; the emails were handwritten.
Other pain points: social media and website enquiries, and the inspection calendar.
What prompted him to ask was hearing friends rave about using AI to automate whole swathes of their businesses while they hit the surf.
I asked Claude how we could help. “Don’t use AI” was its surprising response. Don’t try to fully automate these processes, it continued – that won’t be worth the setup and monitoring costs. All these problems are amenable to sub-AI solutions.
The scheduling problem, Claude thought, was crying out for a shared Google Calendar. A chatbot for social media enquiries would be a sledgehammer for a business this size and would still need human oversight. The monthly reporting issue could be handled by existing off-the shelf software like SafetyCulture that would read inspectors’ site notes and insert them into a report template and mail scheduler.
These responses underlined the truth that AI is still not intuitive or smart enough for you to simply say, for example, “this is my business, deal with social media enquiries”.
It will get that smart, though, and soon. Here’s how AI catches up: two years ago I was using Gemini for legal research and wanted to search my emails for relevant documents. I’d exhausted keywords and was trying to build a knowledge base from my Gmail archive with a bespoke LLM I could query in natural language. It could absolutely be done, but had already taken weeks and was above my technical pay grade – frustrating and time-consuming.
AI chops
Then one day I woke up and presto, Google had integrated Gemini into Gmail. It did exactly what I wanted, with no further effort from me.
My friend’s business problems will likewise be non-issues one day. What he should do in the meantime – this is my tuppence worth, not Claude’s – is make sure he is competent enough with AI so, on that day, he is ready.
This happens by using AI in daily life. And my friend’s monthly report headache is an ideal problem to develop his AI chops. Far better that he use AI to find a non-AI solution than me because of course he knows his business better, and will benefit more from the process.
I suggested he get the $20-a-month (B650) paid for version of Gemini. The paid-for versions of the top models are streets ahead of the free ones. Crucially, they better remember your past conversations, so you can really build on your interactions with it.
Claude is actually a better model currently than Gemini, but Gemini integrates seamlessly with other free apps my friend might use, like Gmail, Drive, Docs and Calendar.
He should have a frank dialogue with Gemini about his problem. He should say, “I’m going to ask for your help to improve monthly reporting at my company. I’m new to AI so you’ll also need to tell me how I should instruct you.” My friend would then explain his business and his problem as best he can, and ask, “what further information do you need from me to give me the best responses?”
Non-AI solutions are still optimal for many business problems, but do use AI to find out what they are.
Joe Smith is Founder of the AI consultancy 2Sigma Consultants. He studied AI at Imperial College Business School and is researching AI’s effects on cognition at Chulalongkorn University. He is author of The Optimized Marketer, a book on how to use AI to promote your business and yourself. Contact: joe@2Sigmaconsultants.com.