I keep seeing posts selling prompt templates like they are the end of consulting, legal work, and half the professional services economy.
“Use this prompt and replace a $500K consultant.” “Goodbye MBB.” “Goodbye lawyers.”
I get why these posts spread. They’re catchy.
But I think they also confuse people about what is actually happening when you prompt a general-purpose LLM.
Just because you say “act like a top-tier consultant” or “act like a lawyer” in a prompt does not mean the model is actually performing the real workflow behind that profession. That’s the part people skip.
To be clear, if you’re using prompts for quick thinking, role play, or exploring ideas, it’s actually great. I do that all the time. But if you’re relying on a simple prompt template for something high-stakes or decision-critical, that’s where you need to be much more cautious about what the system is actually doing.
Professional services are not just a tone of voice or a polished answer. They depend on structured workflows, domain judgment, specialized tools, trusted sources, collaboration across roles, and different methods at different stages of the work.
Take legal work as one example. Real legal work is not just asking AI to sound like an attorney. It involves research on trusted databases, document analysis, drafting standards, confidentiality obligations, verification, and professional review. Even the legal industry’s own AI guidance focuses on exactly these issues, not on “just write a better prompt.”
So when I see “prompt books” marketed as replacements for entire professions, I’m skeptical. Without the right tools, context management, workflow design, evaluation, and agent harness design, many of these systems are still closer to role-play than to real professional execution.
That doesn’t mean AI is not transformative.
It just means the real opportunity is much harder, and much more interesting.
The path forward is not one magical prompt. It’s building agentic systems that combine strong models with the right tools, context, workflow design, and close collaboration with domain experts.
This is also the problem we’ve been working on at NitroLens AI, trying to build what we think of as an “intelligence layer” for strategic analysis. It’s a lot harder than it sounds. That is how AI starts to produce something closer to a real professional-service outcome.
And that work is messy. It takes iteration, testing, evaluation, and deep understanding of how experts actually do their jobs.
This is probably why “buy my prompt book” is a much easier business model.