A good program is not enough, you have to use it well
It is not enough to know which program serves our needs. You have to use it well too.
Did you know it has a program function?
A dear friend of mine, until I showed him that Claude has a program function, was trying to get by on the plain chat interface for everything.
Many people do this: they know the tool but only use one side of it. It is worth going through everything it offers.
Back to the Harry Potter case
A good prompt is precise. In the video article I described what happened to me: I asked for a Harry Potter example in the prompt and got a magical-world background with sparkles. It didn’t look bad, but it wasn’t what I expected. I designed the prompt wrongly.
The lesson: before you send it, think through exactly what you want.
Any other model can do
Any other model can be good for writing the prompt. Next to Claude, I have ChatGPT or Grok word the prompt.
With the right settings, these prompts can also be taken apart and split into parts inside Claude. That way I can decide which part of the task which language model should handle.
New conversations, switching models, building with intent
I open a new conversation several times, switch between language models, and try to word my requests well.
I pay attention to when I use which language model. I only start building when there is really a concrete idea and a plan for it.
The three most important things
- Choosing the right programKnow which program is for what, and don’t only use one side of it.
- Using the program wellA precise prompt, a new conversation when needed, and building with intent.
- Switching well between language modelsPick the right model for each part of the task.
What to watch out for
- A precise prompt: think through what you want before you send it. A bad prompt gives a bad result.
- A new conversation: don’t let one conversation get too long. Open a new one more than once, and word the request well.
- Plan first: only start building once you have a concrete idea and a plan.
Frequently asked questions
Why isn’t a good AI program enough?
It is not enough to know which program serves our needs. You have to use it well: with a good prompt, a good choice of model and building with intent.
What is Claude’s program function good for?
A friend of mine tried to get by on the plain chat interface for everything until I showed him that Claude also has a program function. It is worth getting to know what it offers.
Can I have another AI write the prompt?
Yes. Next to Claude, I have ChatGPT or Grok word the prompt. With the right settings, these can be taken apart and split into parts inside Claude too.
When should I open a new conversation?
I open a new one several times, and switch between language models along the way. I try to word the requests well.
How do I save my quota?
I pay attention to when I use which language model, and only start building when there is a concrete idea and a plan.
ChatGPT, Claude or Grok? My own experience →
Get in touch and we’ll show you how to get more out of AI programs.