Recently, I learned about CRAFTSS: context, reason, audience, format, topic scope, source, soul. It’s a prompting framework. What caught my attention was the soul part. I hadn’t seen that before. Also interesting: it doesn’t include role, which most other frameworks do.
I’ve encountered quite a few of these frameworks now. There’s ROCC (role, objective, context, constraint). There’s CRIT (context, role, interview, task), which I picked up from The AI-Driven Leader. And now CRAFTSS.
Each time I learn a new one, I notice the same pattern: someone figured out what worked for them, used it repeatedly with good results, and shared it. But often the why gets stripped out. They just say, “This is the best way to do it.”
When the Framework Doesn’t Fit
When I first heard about ROCC, I got stuck on the first step. Start with the role? I didn’t always know what role I needed the AI to play. So I couldn’t prompt, because I didn’t know the role.
Then I realized: if I don’t know the role, I can just say that. I can explain what I’m trying to achieve, ask the AI to help me figure out what role would be useful, and go from there.
I actually did this last year when launching my newsletter. I wasn’t sure whether I needed a marketing specialist, a content creator, or something else entirely. So I explained the goal and why it mattered. The AI suggested a few roles with explanations, which helped me choose. After that, my project instructions included something like you are [role].
The key insight: if I don’t know the role, I shouldn’t get stuck. I can approach it differently.
Tools, Not Rules
What works for one person may not work for me, given how I think through problems and how I articulate my thinking. I need to determine what works for me and have multiple approaches available. These are tools, not rules.
I’ve heard of at least one person framing prompts like user stories: as a [person], I want to [something], so that [outcome]. That got me thinking. Personally, I prefer the format: in order to [why], as a [who], I want to [what]. Before I was prompting AI, that’s how I prompted people, asking those questions to develop thoughts and arrive at clarity.
Whether someone uses “as a / I want to / so that” or “in order to / as a / I want to” doesn’t really matter. The point is: are we having the conversations? Are we asking the right questions? Are we achieving clarity? If yes, the format is secondary.
For me, starting with in order to helps because it starts with the why, which is the most important thing. I apply that same logic to my prompts: does my prompt include why I’m asking, or why the task ultimately matters? And if I don’t know what role would help, how can I prompt to discover that role?
How I Prefer Others to Share
I’m more drawn to authors and speakers who present their frameworks as based on personal experience, valid at a particular point in time, and open to evolution. Ideally, they also share things they tried that didn’t work, because what didn’t work for them might work for someone else. That’s worth sharing, too.
What I don’t appreciate is the tone of “if you’re not doing it this way, you’re doing it wrong.” That kind of communication doesn’t help anyone.
Experiments I Want to Run
A few things I want to try.
First, prompting AI with a problem statement, or with what’s on my mind, and having it help me get to the root of the problem, unveiling the need behind it. This could connect to my aspiration/need/problem/solution framework. From there, I could have AI help determine what the prompt or instructions should be, expressed through the story format I prefer. That’s ultimately what I’m trying to achieve, except instead of building software, I’m interacting with AI.
Second, I want to experiment with improv techniques, specifically yes, and, applied to prompting. Three perspectives to try:
-
Yes, but (the contrarian): Yes, you can do that, but it won’t work…
-
Yes, and (building on what I gave it): Yes, and we can make it great by doing the following…
-
Yes, because (justifying through its understanding): Yes, that will be great because…
These can be playful and exploratory, but they’re also another technique I can teach people: a way to unveil values, aspirations, needs, problems, and solutions through conversation.
Trying It Out
I decided to test these improv perspectives on this post itself. I asked AI to respond to my “tools, not rules” idea using each approach.
Yes, but pushed back: Yes, frameworks are tools, but without some structure, people get paralyzed by infinite options. Sometimes a rule helps you start.
That’s fair. The contrarian view revealed something I hadn’t considered: frameworks can reduce decision fatigue. For beginners, having one approach to follow might be better than choosing from many.
Yes, and built on it: Yes, frameworks are tools, and you could create a personal toolkit where you document which approach works best for different types of prompts. Over time, you’d have your own meta-framework.
This expanded the idea. Instead of just using different frameworks situationally, I could track patterns in what works when. That’s actually useful.
Yes, because justified it: Yes, frameworks are tools because the goal is clarity and results, not adherence to a method. The framework that helps you think clearly in the moment is the right one.
This articulated the underlying principle. It’s not about the framework itself, it’s about what the framework enables.
Each perspective surfaced something different. The contrarian helped me see limitations. The builder suggested next steps. The justifier clarified the core value. All three made the idea stronger.
What prompting approach works for you?
Leave a Reply