Before you begin
Open the existing project so your instruction is grounded in the current app.
Write the instruction from the user’s perspective, then state the exact result you want to inspect.
Write the instruction
“Use agents, RAG, a vector database, and the best model” describes technologies.
It does not describe the job.
A stronger prompt says
“A support manager should be able to ask a question, receive an answer based only on approved help articles, see the source used, and route unanswered questions to a person.”
That sentence tells us
Who is acting
What they are trying to do
Which information is allowed
What the result must include
What happens when the system is uncertain
Empromptu AI can help choose the technical approach after the work is clear.
Use technical requirements when they are real constraints, such as
Must run in our cloud
Must use our existing database
Must never send regulated data to an unapproved service
Must expose an API to our product
Start with the job. Add the constraints. Let the architecture serve both.
Common mistakes
Naming tools such as RAG, agents, or vector databases without explaining the business task.
Asking the system to choose the business process for you.
Optimizing technology before the desired result is defined.
Frequently asked questions
How long should a prompt be?
Long enough to define the user, job, input, rules, output, and important constraints. Clear structure matters more than word count.
Can I make several changes in one prompt?
Small related changes can be combined, but changes to different systems are easier to test and recover when requested separately.