In a previous post I discussed few-shot prompting with one example – below some info by blog.coupler.io. (What is Coupler.io? Coupler.io is an sales, finance, and marketing reporting tool that synchronises data between various services on a schedule. © Railsware Products Studio, LLC (from bottom of their page in main page))
Direct instruction vs example based learning (few or one shot prompting)
coupler.io created a customer inquiry response examples and they state that this methodology (few shot prompting) is best used for 3 scenarios:
- Weekly report summaries across sales pipelines or marketing assets
- QA tagging for support chats
- Translating business feedback into product ticket
As we start to use AI for more functions we have to understand the limitations and proper uses (or more effective uses?). The more efficient the process of Q&A gets the more likely errors do not creep into the responses.
And as we create summaries from weekly reports – into automated functions we need to make sure that errors are not included.
So how do we actually do this?
This is a light hearted cartoon I found at LinkedIn, where it states we really should know a little about AI even if we really want to run it now.
Questioning frameworks that work
In prompting, the way you frame a question directly influences the model’s logic, focus, and creativity.
The frameworks below are not shortcuts; they are mental models that work across analytics, copywriting, product strategy, and executive decision-making.
We are of course training the AI in how we want the answers to be shaped… so if we can give an example (like jaguar branding rework which failed and the examples we wish to direct to).
Remember the model needs positive and negative examples.
To help with communication to others also interested in doing AI instructions and more… Setting up a discord server fixvirus-server to experiment with some Zapier workflows .
Discord server is by invitation only so users are not getting mobbed by others (contact me and I will vet for now. Zapier will help me manage the Discord server, and create some automation – once it is working better I will post what I can about this.
Starting with new tools is interesting and exciting at same time 🙂
Will work through the challenges – it is supposed to be a forum to discuss AI -and cybersecurity
The issue is one must know how something works to do a good test on it, so if testing AI processes one must know the setup and uses. Thus will set up “AI” processes.
Including the funny What do we want image – since it encapsulates exactly what I have been working with and understanding about new technologies – it is good to know about the new tech before fully implementing, as there are likely negatives that will creep into the business.
Contact me to discuss what and how you want to use AI in your business.
One does not want to deny AI completely, (A NO to all AI) – instead think about how to pilot it and work on the security – to ensure transparency:
The idea is to implement secure systems with continuous improvement!!.
Must have self assessments as well as outside audits.
There is an aspect of AI governance – which is where to put AI must be pushed from above – (CEO or other heavy hitters in organization) as the potential is there to make significant changes. Although security needs to be built in from beginning.
FYI – I fed invideo.io the following: (just like in cartoon above)
When CEOs want AI but don’t know why — a sharp 45-second LinkedIn-ready clip for startup teams.
This tongue-in-cheek short follows the script word-for-word:
one CEO asks “Who are we?” (CEOs),
“What do we want?” (AI), “AI, to do what?” (We don’t know!),
“When do we want it?” (We want it now).
The rest the AI in invideo wanted to add this for the fun of it…
Perfect for sparking conversations about leadership, product strategy, and AI adoption in startups. Use in meetings, pitch decks, or social posts to highlight urgency vs. clarity in AI decisions. Like and share if this hit home — start the discussion in your team. #AI #CEOs #Startups #Leadership #ProductStrategy #LinkedInVideo #Shorts
youtube video where I uploaded the short video:
My point is to hopefully kindle a governance conversation with the leaders in companies.