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
An app cannot reliably use information that has not been identified.
Tell Empromptu AI where the knowledge comes from
Documents or PDFs
A spreadsheet or CSV
A database
An API
Approved help articles
User-entered information
An existing codebase
A combination of sources
Then identify the source of truth.
Example
“The app receives an onboarding request from ServiceNow. Workday is the source of truth for location and department. Our access matrix contains the rules. If the systems disagree, do not guess. Show the conflict for approval.”
That last sentence is often more important than the data connection itself. It explains which system wins and what happens when the information is incomplete.
Start with sample or synthetic data when the real data is sensitive. Connect production data only after the workflow behaves correctly.
Common mistakes
Connecting every available data source at once.
Using production or regulated data before controls are ready.
Failing to name the approved source of truth.
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.